{"version":"1.0.0","segments":[{"startTime":0.3,"endTime":3.12,"body":"- [Announcer] Welcome to"},{"startTime":0.3,"endTime":3.12,"body":"this Mtech Access webinar."},{"startTime":3.12,"endTime":6.24,"body":"At Mtech Access, we"},{"startTime":3.12,"endTime":6.24,"body":"provide health economics"},{"startTime":6.24,"endTime":7.53,"body":"and outcomes research"},{"startTime":7.53,"endTime":9.45,"body":"and market access services"},{"startTime":9.45,"endTime":12.0,"body":"from strategy through to implementation."},{"startTime":12.0,"endTime":16.08,"body":"Get in touch today to discuss"},{"startTime":12.0,"endTime":16.08,"body":"your market access goals."},{"startTime":16.08,"endTime":18.513,"body":"First, though, I hope"},{"startTime":16.08,"endTime":18.513,"body":"you enjoy the webinar."},{"startTime":19.86,"endTime":22.14,"body":"- Hello everybody and"},{"startTime":19.86,"endTime":22.14,"body":"welcome to this webinar."},{"startTime":22.14,"endTime":25.23,"body":"Thank you all very much for joining today."},{"startTime":25.23,"endTime":28.47,"body":"The topic of our webinar"},{"startTime":25.23,"endTime":28.47,"body":"today is \"Demystifying Health"},{"startTime":28.47,"endTime":31.524,"body":"Economic Model Classifications"},{"startTime":28.47,"endTime":31.524,"body":"and Structures.\""},{"startTime":31.524,"endTime":34.44,"body":"This will include discussing cohort"},{"startTime":34.44,"endTime":36.33,"body":"versus patient-level approaches"},{"startTime":36.33,"endTime":38.49,"body":"and we will explain the key strengths"},{"startTime":38.49,"endTime":41.82,"body":"and limitations of six health"},{"startTime":38.49,"endTime":41.82,"body":"economic model structures,"},{"startTime":41.82,"endTime":45.393,"body":"which can be used to support"},{"startTime":41.82,"endTime":45.393,"body":"market access activities."},{"startTime":50.55,"endTime":53.91,"body":"So to start off with, we"},{"startTime":50.55,"endTime":53.91,"body":"will introduce ourselves."},{"startTime":53.91,"endTime":55.08,"body":"There are two of us here today"},{"startTime":55.08,"endTime":58.29,"body":"from the Health Economics"},{"startTime":55.08,"endTime":58.29,"body":"team at Mtech Access."},{"startTime":58.29,"endTime":60.12,"body":"So my name is Hannah Gillies,"},{"startTime":60.12,"endTime":61.802,"body":"I'm a Consultant Health Economist"},{"startTime":61.802,"endTime":65.344,"body":"and I've been working in"},{"startTime":61.802,"endTime":65.344,"body":"market access since 2012,"},{"startTime":65.344,"endTime":68.19,"body":"designing and developing economic models"},{"startTime":68.19,"endTime":71.13,"body":"across a wide range of disease areas."},{"startTime":71.13,"endTime":73.622,"body":"I have experience in modelling"},{"startTime":73.622,"endTime":75.48,"body":"with many different model structures"},{"startTime":75.48,"endTime":78.57,"body":"including Markov models,"},{"startTime":75.48,"endTime":78.57,"body":"discrete event simulations"},{"startTime":78.57,"endTime":80.82,"body":"and partitioned survival models,"},{"startTime":80.82,"endTime":84.51,"body":"as well as budget impact and"},{"startTime":80.82,"endTime":84.51,"body":"peer communication tools."},{"startTime":84.51,"endTime":87.183,"body":"And I'm joined today"},{"startTime":84.51,"endTime":87.183,"body":"by my colleague Danny."},{"startTime":88.229,"endTime":91.646,"body":"- Hi everyone and welcome to the webinar."},{"startTime":91.646,"endTime":93.5,"body":"My name's Danny MacDonald,"},{"startTime":93.5,"endTime":96.379,"body":"I'm an Associate Consultant here at Mtech."},{"startTime":96.379,"endTime":98.25,"body":"A bit of background about me."},{"startTime":98.25,"endTime":100.704,"body":"So I've got roughly six and a"},{"startTime":98.25,"endTime":100.704,"body":"half years worth of experience"},{"startTime":100.704,"endTime":102.42,"body":"as a health economist"},{"startTime":102.42,"endTime":104.905,"body":"and that covers a broad"},{"startTime":102.42,"endTime":104.905,"body":"range of disease areas."},{"startTime":104.905,"endTime":107.67,"body":"I've worked on several HTA submissions"},{"startTime":107.67,"endTime":109.927,"body":"covering oncology and non-oncology."},{"startTime":109.927,"endTime":113.94,"body":"These include global"},{"startTime":109.927,"endTime":113.94,"body":"models, submissions to NICE"},{"startTime":113.94,"endTime":116.22,"body":"and CADITH, so in Canada."},{"startTime":116.22,"endTime":118.41,"body":"My modelling experience covers a lot"},{"startTime":118.41,"endTime":121.5,"body":"of cost effectiveness"},{"startTime":118.41,"endTime":121.5,"body":"models, budget impact models,"},{"startTime":121.5,"endTime":123.162,"body":"and also statistical analysis"},{"startTime":123.162,"endTime":126.298,"body":"so such as discrete choice experiments"},{"startTime":126.298,"endTime":130.23,"body":"and surrogacy endpoint analysis."},{"startTime":130.23,"endTime":132.81,"body":"So it's a pleasure to"},{"startTime":130.23,"endTime":132.81,"body":"have you all with us today"},{"startTime":132.81,"endTime":134.121,"body":"for the webinar."},{"startTime":134.121,"endTime":135.57,"body":"And Hannah, if you'd be able"},{"startTime":135.57,"endTime":137.85,"body":"to move on to the next slide please?"},{"startTime":137.85,"endTime":138.81,"body":"So a bit of background"},{"startTime":138.81,"endTime":140.885,"body":"about what we're going to"},{"startTime":138.81,"endTime":140.885,"body":"be talking about today."},{"startTime":140.885,"endTime":143.88,"body":"So we're going to start off"},{"startTime":140.885,"endTime":143.88,"body":"by providing an introduction"},{"startTime":143.88,"endTime":145.47,"body":"to health economic evaluation."},{"startTime":145.47,"endTime":148.71,"body":"So we're very much going"},{"startTime":145.47,"endTime":148.71,"body":"to be setting the scene,"},{"startTime":148.71,"endTime":150.84,"body":"so starting sort of building blocks,"},{"startTime":150.84,"endTime":153.365,"body":"providing a definition about"},{"startTime":150.84,"endTime":153.365,"body":"health economic evaluation"},{"startTime":153.365,"endTime":156.48,"body":"and then we're going to move into sort"},{"startTime":156.48,"endTime":157.89,"body":"of a real world application."},{"startTime":157.89,"endTime":160.083,"body":"How is it applied across the world?"},{"startTime":161.04,"endTime":162.45,"body":"What are the key considerations"},{"startTime":162.45,"endTime":165.87,"body":"when applying health economic evaluation?"},{"startTime":165.87,"endTime":168.427,"body":"We're then going to look at"},{"startTime":165.87,"endTime":168.427,"body":"specific modelling approaches."},{"startTime":168.427,"endTime":171.81,"body":"So there's two sort of key distinct types"},{"startTime":171.81,"endTime":173.04,"body":"of modelling approaches here."},{"startTime":173.04,"endTime":175.08,"body":"So you've got cohort models"},{"startTime":175.08,"endTime":177.18,"body":"and you've got patient-level models."},{"startTime":177.18,"endTime":178.85,"body":"There's also a little"},{"startTime":177.18,"endTime":178.85,"body":"bit of overlap there."},{"startTime":178.85,"endTime":179.97,"body":"So what we're going to be doing"},{"startTime":179.97,"endTime":183.03,"body":"is taking six specific"},{"startTime":179.97,"endTime":183.03,"body":"modelling approaches"},{"startTime":183.03,"endTime":185.497,"body":"and discussing them in a bit more detail."},{"startTime":185.497,"endTime":188.82,"body":"We're also going to be"},{"startTime":185.497,"endTime":188.82,"body":"looking at hybrid modelling."},{"startTime":188.82,"endTime":191.25,"body":"So that's when you merge"},{"startTime":188.82,"endTime":191.25,"body":"two models together"},{"startTime":191.25,"endTime":194.07,"body":"when a specific situation requires it."},{"startTime":194.07,"endTime":195.42,"body":"We'll then look at a checklist"},{"startTime":195.42,"endTime":197.4,"body":"for the choice of model structure,"},{"startTime":197.4,"endTime":200.25,"body":"which have sort of a list"},{"startTime":197.4,"endTime":200.25,"body":"of key considerations"},{"startTime":200.25,"endTime":203.91,"body":"that we think are useful when"},{"startTime":200.25,"endTime":203.91,"body":"approaching a modelling project"},{"startTime":203.91,"endTime":206.58,"body":"or a model sort of decision problem."},{"startTime":206.58,"endTime":209.1,"body":"We'll then summarise our key findings"},{"startTime":209.1,"endTime":210.48,"body":"and then there'll be a section at the end"},{"startTime":210.48,"endTime":214.35,"body":"where we'll be responding"},{"startTime":210.48,"endTime":214.35,"body":"to questions at the end."},{"startTime":214.35,"endTime":216.01,"body":"So next slide please, Hannah."},{"startTime":218.67,"endTime":223.62,"body":"So to start off, why is health"},{"startTime":218.67,"endTime":223.62,"body":"economic evaluation needed?"},{"startTime":223.62,"endTime":226.74,"body":"Well, the definition of"},{"startTime":223.62,"endTime":226.74,"body":"health economic evaluation is"},{"startTime":226.74,"endTime":229.71,"body":"that it's the comparison"},{"startTime":226.74,"endTime":229.71,"body":"of alternative options"},{"startTime":229.71,"endTime":232.29,"body":"in terms of their costs and consequences."},{"startTime":232.29,"endTime":236.88,"body":"So consequences in this"},{"startTime":232.29,"endTime":236.88,"body":"instance being health benefits."},{"startTime":236.88,"endTime":239.38,"body":"So what are the main health"},{"startTime":236.88,"endTime":239.38,"body":"benefits and the costs"},{"startTime":240.324,"endTime":241.8,"body":"and how are we able to compare them?"},{"startTime":241.8,"endTime":243.458,"body":"So if we can see on the side here,"},{"startTime":243.458,"endTime":245.1,"body":"health economic evaluations,"},{"startTime":245.1,"endTime":246.9,"body":"it involves appraising the health benefits"},{"startTime":246.9,"endTime":248.97,"body":"that can be gained from an intervention"},{"startTime":248.97,"endTime":251.88,"body":"and evaluating whether the"},{"startTime":248.97,"endTime":251.88,"body":"additional gains provide value"},{"startTime":251.88,"endTime":254.877,"body":"for money when compared the"},{"startTime":251.88,"endTime":254.877,"body":"best available alternative"},{"startTime":254.877,"endTime":257.97,"body":"for the use of a cost effectiveness model."},{"startTime":257.97,"endTime":261.507,"body":"So health economic evaluation"},{"startTime":257.97,"endTime":261.507,"body":"is crucial for decision making"},{"startTime":261.507,"endTime":265.94,"body":"and it allows for this, it"},{"startTime":261.507,"endTime":265.94,"body":"allows health agencies to decide"},{"startTime":265.94,"endTime":269.49,"body":"what is the most efficient"},{"startTime":265.94,"endTime":269.49,"body":"allocation of money."},{"startTime":269.49,"endTime":270.96,"body":"Which is crucial when you've got,"},{"startTime":270.96,"endTime":273.06,"body":"say, a constrained budget system,"},{"startTime":273.06,"endTime":275.7,"body":"so say in the UK you've"},{"startTime":273.06,"endTime":275.7,"body":"got a certain amount"},{"startTime":275.7,"endTime":277.08,"body":"of healthcare expenditure,"},{"startTime":277.08,"endTime":279.36,"body":"how can we spend that money efficiently?"},{"startTime":279.36,"endTime":281.96,"body":"And that's where health"},{"startTime":279.36,"endTime":281.96,"body":"economic valuation comes in."},{"startTime":283.65,"endTime":286.35,"body":"A willingness-to-pay"},{"startTime":283.65,"endTime":286.35,"body":"threshold is often used"},{"startTime":286.35,"endTime":289.11,"body":"to compare the cost"},{"startTime":286.35,"endTime":289.11,"body":"effectiveness of treatments."},{"startTime":289.11,"endTime":292.77,"body":"So with NICE, we have a"},{"startTime":289.11,"endTime":292.77,"body":"30,000 pound QALY measurement."},{"startTime":292.77,"endTime":294.96,"body":"So a QALY being a"},{"startTime":292.77,"endTime":294.96,"body":"quality-adjusted life year,"},{"startTime":294.96,"endTime":297.36,"body":"which is a measure of health benefit"},{"startTime":297.36,"endTime":300.72,"body":"which takes into account"},{"startTime":297.36,"endTime":300.72,"body":"the additional life years"},{"startTime":300.72,"endTime":303.72,"body":"that can be experienced by"},{"startTime":300.72,"endTime":303.72,"body":"introducing an intervention"},{"startTime":303.72,"endTime":306.78,"body":"and applies a utility value to that"},{"startTime":306.78,"endTime":309.483,"body":"to analyse the quality"},{"startTime":306.78,"endTime":309.483,"body":"of life that we receive."},{"startTime":310.86,"endTime":312.88,"body":"What you do then is you're"},{"startTime":310.86,"endTime":312.88,"body":"able to estimate an ICER,"},{"startTime":312.88,"endTime":315.69,"body":"so an incremental cost"},{"startTime":312.88,"endTime":315.69,"body":"effectiveness ratio,"},{"startTime":315.69,"endTime":317.37,"body":"and assess whether that falls"},{"startTime":317.37,"endTime":319.47,"body":"below the willingness-to-pay threshold."},{"startTime":319.47,"endTime":321.518,"body":"So an ICER is the difference"},{"startTime":321.518,"endTime":324.99,"body":"or it's a formula which looks"},{"startTime":321.518,"endTime":324.99,"body":"at the difference in the cost"},{"startTime":324.99,"endTime":327.78,"body":"between an intervention"},{"startTime":324.99,"endTime":327.78,"body":"and a comparative treatment"},{"startTime":327.78,"endTime":329.79,"body":"and divides that by the difference"},{"startTime":329.79,"endTime":332.22,"body":"in the health benefits of the QALYs."},{"startTime":332.22,"endTime":334.718,"body":"What you'll then able to sort"},{"startTime":334.718,"endTime":336.12,"body":"of generate is a cost"},{"startTime":334.718,"endTime":336.12,"body":"per QALY measurement."},{"startTime":336.12,"endTime":337.724,"body":"So say let's say"},{"startTime":337.724,"endTime":339.63,"body":"we've got a 25,000 pound QALY measurement,"},{"startTime":339.63,"endTime":342.09,"body":"you compare that to the"},{"startTime":339.63,"endTime":342.09,"body":"willingness-to-pay threshold"},{"startTime":342.09,"endTime":343.768,"body":"and if it falls below the threshold"},{"startTime":343.768,"endTime":346.26,"body":"then it's deemed as cost effective."},{"startTime":346.26,"endTime":347.721,"body":"If it falls above the threshold"},{"startTime":347.721,"endTime":351.003,"body":"then it is deemed as not"},{"startTime":347.721,"endTime":351.003,"body":"being cost effective."},{"startTime":352.26,"endTime":354.03,"body":"If we look at the righthand side here,"},{"startTime":354.03,"endTime":356.704,"body":"the side there are three"},{"startTime":354.03,"endTime":356.704,"body":"sort of key areas where cost"},{"startTime":356.704,"endTime":360.66,"body":"or where health economic"},{"startTime":356.704,"endTime":360.66,"body":"evaluation is specifically needed."},{"startTime":360.66,"endTime":363.719,"body":"So as mentioned, there's"},{"startTime":360.66,"endTime":363.719,"body":"increased pressure on budgets,"},{"startTime":363.719,"endTime":366.12,"body":"so justification of expenditure"},{"startTime":366.12,"endTime":368.52,"body":"within the healthcare"},{"startTime":366.12,"endTime":368.52,"body":"system is very important."},{"startTime":368.52,"endTime":371.44,"body":"It allows you to assist"},{"startTime":368.52,"endTime":371.44,"body":"decision-making bodies."},{"startTime":371.44,"endTime":373.98,"body":"So it provides a quantitative framework"},{"startTime":373.98,"endTime":375.781,"body":"for evaluating technologies."},{"startTime":375.781,"endTime":378.0,"body":"It provides a number in addition"},{"startTime":378.0,"endTime":380.91,"body":"to the qualitative submission"},{"startTime":380.91,"endTime":383.55,"body":"that might be provided by"},{"startTime":380.91,"endTime":383.55,"body":"a pharmaceutical company."},{"startTime":383.55,"endTime":385.77,"body":"And this very much gives"},{"startTime":383.55,"endTime":385.77,"body":"you sort of a data point"},{"startTime":385.77,"endTime":387.6,"body":"from which you can base"},{"startTime":385.77,"endTime":387.6,"body":"several discussions"},{"startTime":387.6,"endTime":388.83,"body":"and then explore reasons"},{"startTime":388.83,"endTime":393.54,"body":"for why the cost effectiveness"},{"startTime":388.83,"endTime":393.54,"body":"value's being reached."},{"startTime":393.54,"endTime":396.3,"body":"It's also allows to support interventions."},{"startTime":396.3,"endTime":397.65,"body":"So a robust model"},{"startTime":397.65,"endTime":400.74,"body":"that can describe a real"},{"startTime":397.65,"endTime":400.74,"body":"world setting can be crucial"},{"startTime":400.74,"endTime":402.502,"body":"in terms of A HTA submission."},{"startTime":402.502,"endTime":404.37,"body":"It can support an argument"},{"startTime":404.37,"endTime":406.86,"body":"or it can generate an argument"},{"startTime":404.37,"endTime":406.86,"body":"which can then be submitted"},{"startTime":406.86,"endTime":410.67,"body":"to the health agency and"},{"startTime":406.86,"endTime":410.67,"body":"it can very much frame"},{"startTime":410.67,"endTime":411.744,"body":"and begin a discussion"},{"startTime":411.744,"endTime":414.933,"body":"for why a technology should be approved."},{"startTime":415.98,"endTime":417.603,"body":"The next slide please, Hannah."},{"startTime":419.46,"endTime":422.22,"body":"So around the world, health"},{"startTime":419.46,"endTime":422.22,"body":"economic evaluation is a key"},{"startTime":422.22,"endTime":423.603,"body":"and crucial component."},{"startTime":424.44,"endTime":425.88,"body":"As you can see here on the slide,"},{"startTime":425.88,"endTime":428.58,"body":"there are five sort of"},{"startTime":425.88,"endTime":428.58,"body":"areas we've called out"},{"startTime":428.58,"endTime":432.03,"body":"of different ways that economic"},{"startTime":428.58,"endTime":432.03,"body":"evaluation can be used."},{"startTime":432.03,"endTime":435.03,"body":"So I'll start especially for"},{"startTime":432.03,"endTime":435.03,"body":"those that's on audio readout."},{"startTime":435.03,"endTime":438.498,"body":"So we've got a budget-impact"},{"startTime":435.03,"endTime":438.498,"body":"focused approach."},{"startTime":438.498,"endTime":441.879,"body":"So often when you are"},{"startTime":438.498,"endTime":441.879,"body":"performing economic evaluation,"},{"startTime":441.879,"endTime":444.915,"body":"at least with NICE in particular,"},{"startTime":444.915,"endTime":447.15,"body":"you look at a cost effectiveness model"},{"startTime":447.15,"endTime":449.18,"body":"and you also look at"},{"startTime":447.15,"endTime":449.18,"body":"a budget impact model."},{"startTime":449.18,"endTime":452.31,"body":"The cost effectiveness model"},{"startTime":449.18,"endTime":452.31,"body":"usually takes more precedence"},{"startTime":452.31,"endTime":453.51,"body":"over the budget impacts model,"},{"startTime":453.51,"endTime":455.01,"body":"but it's still there as an analysis"},{"startTime":455.01,"endTime":457.53,"body":"of importance for health agencies."},{"startTime":457.53,"endTime":461.28,"body":"In some markets the budget"},{"startTime":457.53,"endTime":461.28,"body":"impact aspect has more prominence"},{"startTime":461.28,"endTime":462.718,"body":"and it's of more importance,"},{"startTime":462.718,"endTime":465.063,"body":"say than the cost effectiveness output."},{"startTime":465.96,"endTime":469.29,"body":"So we can see here markets such as Italy"},{"startTime":469.29,"endTime":472.865,"body":"and Spain, for example,"},{"startTime":469.29,"endTime":472.865,"body":"specifically look at those areas."},{"startTime":472.865,"endTime":475.078,"body":"There's patient led approaches,"},{"startTime":475.078,"endTime":479.16,"body":"so that's in particular looking"},{"startTime":475.078,"endTime":479.16,"body":"at Indonesia and Vietnam"},{"startTime":479.16,"endTime":482.039,"body":"which look at patient"},{"startTime":479.16,"endTime":482.039,"body":"sort of level outcomes"},{"startTime":482.039,"endTime":484.05,"body":"or patient focused outcomes."},{"startTime":484.05,"endTime":486.15,"body":"And that's incorporated as a sort"},{"startTime":486.15,"endTime":488.35,"body":"of crucial point in the"},{"startTime":486.15,"endTime":488.35,"body":"economic evaluation."},{"startTime":489.87,"endTime":492.72,"body":"There's then health-economic"},{"startTime":489.87,"endTime":492.72,"body":"or cost effectiveness-focused."},{"startTime":492.72,"endTime":494.643,"body":"So the main example being NICE."},{"startTime":495.87,"endTime":497.43,"body":"So in these instances,"},{"startTime":497.43,"endTime":500.149,"body":"a cost effectiveness approach is crucial"},{"startTime":500.149,"endTime":504.672,"body":"and is sort of set up as"},{"startTime":500.149,"endTime":504.672,"body":"the main modelling approach"},{"startTime":504.672,"endTime":507.39,"body":"when looking at economic evaluation."},{"startTime":507.39,"endTime":508.41,"body":"So as mentioned before,"},{"startTime":508.41,"endTime":511.5,"body":"looking to achieve cost"},{"startTime":508.41,"endTime":511.5,"body":"effectiveness with the ICER"},{"startTime":511.5,"endTime":513.472,"body":"relative to a"},{"startTime":511.5,"endTime":513.472,"body":"willingness-to-pay threshold."},{"startTime":513.472,"endTime":515.91,"body":"There's then clinically focused countries."},{"startTime":515.91,"endTime":519.03,"body":"So the main example here"},{"startTime":515.91,"endTime":519.03,"body":"I'd say is probably France."},{"startTime":519.03,"endTime":521.88,"body":"So with a France, or French market,"},{"startTime":521.88,"endTime":525.893,"body":"you've got the ASMR scale"},{"startTime":521.88,"endTime":525.893,"body":"which ranks drugs compared"},{"startTime":525.893,"endTime":529.74,"body":"to existing treatment options"},{"startTime":525.893,"endTime":529.74,"body":"from a clinical point of view."},{"startTime":529.74,"endTime":533.61,"body":"So the first level of ASMR is"},{"startTime":529.74,"endTime":533.61,"body":"a major improvement clinically"},{"startTime":533.61,"endTime":536.354,"body":"and the fifth scale, it's"},{"startTime":533.61,"endTime":536.354,"body":"on a five point scale."},{"startTime":536.354,"endTime":540.09,"body":"The fifth point represents no improvement."},{"startTime":540.09,"endTime":544.11,"body":"And those sort of clinical"},{"startTime":540.09,"endTime":544.11,"body":"rankings there are"},{"startTime":544.11,"endTime":545.4,"body":"of specific importance"},{"startTime":545.4,"endTime":547.738,"body":"when considering the economic"},{"startTime":545.4,"endTime":547.738,"body":"evaluation component."},{"startTime":547.738,"endTime":549.137,"body":"And then in America,"},{"startTime":549.137,"endTime":552.375,"body":"there's the competitive"},{"startTime":549.137,"endTime":552.375,"body":"insurance-based model,"},{"startTime":552.375,"endTime":554.91,"body":"which I'm sure we're all familiar with"},{"startTime":554.91,"endTime":557.519,"body":"when we've heard it around the news."},{"startTime":557.519,"endTime":560.85,"body":"It's very much sort of"},{"startTime":557.519,"endTime":560.85,"body":"an insurer-focused model"},{"startTime":560.85,"endTime":562.796,"body":"in terms of a price negotiations."},{"startTime":562.796,"endTime":565.95,"body":"There's probably a"},{"startTime":562.796,"endTime":565.95,"body":"webinar in itself talking"},{"startTime":565.95,"endTime":567.75,"body":"about the insurance-based model,"},{"startTime":567.75,"endTime":570.3,"body":"but it's very unique to the United States."},{"startTime":570.3,"endTime":573.297,"body":"And yeah, it's a very"},{"startTime":570.3,"endTime":573.297,"body":"interesting one to go into"},{"startTime":573.297,"endTime":577.41,"body":"as yeah, I recommend if you have a look."},{"startTime":577.41,"endTime":579.01,"body":"Okay, next slide please, Hannah."},{"startTime":581.73,"endTime":583.581,"body":"So if we bring it back to modelling"},{"startTime":583.581,"endTime":586.56,"body":"and we want to provide a top line overview"},{"startTime":586.56,"endTime":589.458,"body":"of why modelling is important"},{"startTime":586.56,"endTime":589.458,"body":"in economic evaluation,"},{"startTime":589.458,"endTime":594.03,"body":"I think I've displayed"},{"startTime":589.458,"endTime":594.03,"body":"it here in this slide"},{"startTime":594.03,"endTime":595.95,"body":"and I'll provide a note for you."},{"startTime":595.95,"endTime":597.36,"body":"So cost effectiveness modelling,"},{"startTime":597.36,"endTime":600.03,"body":"it underpins health economic evaluation"},{"startTime":600.03,"endTime":603.42,"body":"providing an analytical"},{"startTime":600.03,"endTime":603.42,"body":"framework to evaluate data"},{"startTime":603.42,"endTime":605.55,"body":"and produce a quantitative outcome"},{"startTime":605.55,"endTime":607.38,"body":"indicating the cost effectiveness"},{"startTime":607.38,"endTime":610.68,"body":"of an intervention treatment"},{"startTime":607.38,"endTime":610.68,"body":"against a relative comparator."},{"startTime":610.68,"endTime":613.14,"body":"So there's three key"},{"startTime":610.68,"endTime":613.14,"body":"areas of consideration"},{"startTime":613.14,"endTime":615.51,"body":"when you are conceptualising a model"},{"startTime":615.51,"endTime":617.97,"body":"or when you're looking to"},{"startTime":615.51,"endTime":617.97,"body":"approach a modelling project."},{"startTime":617.97,"endTime":620.7,"body":"So we've got the model"},{"startTime":617.97,"endTime":620.7,"body":"inputs, the model structure,"},{"startTime":620.7,"endTime":622.2,"body":"and the model outputs."},{"startTime":622.2,"endTime":623.79,"body":"So I'm just going to go down them here"},{"startTime":623.79,"endTime":627.21,"body":"and give a brief overview of"},{"startTime":623.79,"endTime":627.21,"body":"key considerations to have."},{"startTime":627.21,"endTime":630.15,"body":"So in terms of model inputs,"},{"startTime":627.21,"endTime":630.15,"body":"I've highlighted four here,"},{"startTime":630.15,"endTime":632.49,"body":"so the first one being efficacy."},{"startTime":632.49,"endTime":634.29,"body":"So this is basically,"},{"startTime":634.29,"endTime":636.954,"body":"how well does the"},{"startTime":634.29,"endTime":636.954,"body":"intervention treatment perform"},{"startTime":636.954,"endTime":639.63,"body":"compared to comparative treatment?"},{"startTime":639.63,"endTime":643.218,"body":"The efficacy data is usually"},{"startTime":639.63,"endTime":643.218,"body":"obtained from a clinical trial"},{"startTime":643.218,"endTime":646.542,"body":"that the pharmaceutical"},{"startTime":643.218,"endTime":646.542,"body":"company may have run."},{"startTime":646.542,"endTime":650.07,"body":"And then comparative data"},{"startTime":646.542,"endTime":650.07,"body":"can either be collected"},{"startTime":650.07,"endTime":653.07,"body":"via the clinical trial if it's"},{"startTime":650.07,"endTime":653.07,"body":"a head-to-head clinical trial"},{"startTime":653.07,"endTime":655.83,"body":"or in some cases if you're"},{"startTime":653.07,"endTime":655.83,"body":"using a single arm trial,"},{"startTime":655.83,"endTime":658.775,"body":"it may have to be obtained"},{"startTime":655.83,"endTime":658.775,"body":"through an indirect measurement."},{"startTime":658.775,"endTime":661.597,"body":"So for example, an indirect"},{"startTime":658.775,"endTime":661.597,"body":"treatment comparison"},{"startTime":661.597,"endTime":665.474,"body":"which takes at the"},{"startTime":661.597,"endTime":665.474,"body":"trial of the comparator,"},{"startTime":665.474,"endTime":667.8,"body":"it tries to establish similarities"},{"startTime":667.8,"endTime":668.97,"body":"in the standard of care arms"},{"startTime":668.97,"endTime":670.59,"body":"that were used in the trial"},{"startTime":670.59,"endTime":672.99,"body":"or it tries to adjust for differences"},{"startTime":672.99,"endTime":675.279,"body":"in the patient populations"},{"startTime":672.99,"endTime":675.279,"body":"between the two trials"},{"startTime":675.279,"endTime":678.54,"body":"and it allows you to"},{"startTime":675.279,"endTime":678.54,"body":"generate various measurements"},{"startTime":678.54,"endTime":680.37,"body":"or ratios which can then be applied"},{"startTime":680.37,"endTime":682.92,"body":"to the intervention arms efficacy"},{"startTime":682.92,"endTime":684.543,"body":"to estimate a comparator arm."},{"startTime":686.7,"endTime":688.58,"body":"The second area here is adverse events"},{"startTime":688.58,"endTime":690.152,"body":"to the safety profile."},{"startTime":690.152,"endTime":693.06,"body":"These are usually collected"},{"startTime":690.152,"endTime":693.06,"body":"from the clinical trial"},{"startTime":693.06,"endTime":695.997,"body":"and in the model there's"},{"startTime":693.06,"endTime":695.997,"body":"usually a threshold"},{"startTime":695.997,"endTime":698.4,"body":"of inclusion that's used."},{"startTime":698.4,"endTime":701.036,"body":"So usually in models it could be 3%."},{"startTime":701.036,"endTime":704.275,"body":"So any grade three for adverse event"},{"startTime":704.275,"endTime":707.82,"body":"that achieves a 3% sort of instance"},{"startTime":707.82,"endTime":710.033,"body":"in the clinical trial would be included"},{"startTime":710.033,"endTime":713.1,"body":"and that will also apply"},{"startTime":710.033,"endTime":713.1,"body":"for comparator arms."},{"startTime":713.1,"endTime":714.78,"body":"Adverse events are also quite important"},{"startTime":714.78,"endTime":717.0,"body":"from a qualitative point of view."},{"startTime":717.0,"endTime":720.33,"body":"If you're able to emphasise"},{"startTime":717.0,"endTime":720.33,"body":"the safety profile benefits"},{"startTime":720.33,"endTime":724.293,"body":"that can be achieved, it can"},{"startTime":720.33,"endTime":724.293,"body":"be beneficial to a submission."},{"startTime":725.37,"endTime":726.66,"body":"There's costs and resource use,"},{"startTime":726.66,"endTime":729.0,"body":"so costs key sources say from the UK"},{"startTime":729.0,"endTime":731.541,"body":"and NICE would be P-S-S-I-U"},{"startTime":729.0,"endTime":731.541,"body":"and NHS reference costs."},{"startTime":731.541,"endTime":734.821,"body":"Resource use are usually"},{"startTime":731.541,"endTime":734.821,"body":"applied in health states."},{"startTime":734.821,"endTime":739.2,"body":"So states, so say for"},{"startTime":734.821,"endTime":739.2,"body":"example a progression state"},{"startTime":739.2,"endTime":740.79,"body":"as we're going to talk about a bit later,"},{"startTime":740.79,"endTime":742.62,"body":"you might have a certain"},{"startTime":740.79,"endTime":742.62,"body":"amount of resource use"},{"startTime":742.62,"endTime":744.96,"body":"and those are accrued in each model cycle."},{"startTime":744.96,"endTime":748.613,"body":"So each duration of time"},{"startTime":744.96,"endTime":748.613,"body":"that a model cycle occurs,"},{"startTime":748.613,"endTime":751.533,"body":"the resources will be"},{"startTime":748.613,"endTime":751.533,"body":"incurred by the patient."},{"startTime":752.73,"endTime":754.207,"body":"Utilities were also consideration here,"},{"startTime":754.207,"endTime":757.68,"body":"these are often collected"},{"startTime":754.207,"endTime":757.68,"body":"in clinical trials"},{"startTime":757.68,"endTime":759.834,"body":"but can also can be picked up"},{"startTime":759.834,"endTime":761.73,"body":"from a systematic literature review"},{"startTime":761.73,"endTime":763.951,"body":"for the indication of interest."},{"startTime":763.951,"endTime":766.475,"body":"So moving now to the model structure,"},{"startTime":766.475,"endTime":768.229,"body":"this is very much the framework"},{"startTime":768.229,"endTime":771.45,"body":"which establishes how the"},{"startTime":768.229,"endTime":771.45,"body":"inputs are synthesised"},{"startTime":771.45,"endTime":773.19,"body":"and turned into outputs."},{"startTime":773.19,"endTime":776.472,"body":"So you're able to get your"},{"startTime":773.19,"endTime":776.472,"body":"key sort of output data point."},{"startTime":776.472,"endTime":778.77,"body":"There's key considerations"},{"startTime":776.472,"endTime":778.77,"body":"for the structure"},{"startTime":778.77,"endTime":780.57,"body":"as we're going to go into."},{"startTime":780.57,"endTime":781.53,"body":"As a top line,"},{"startTime":781.53,"endTime":785.655,"body":"I'd say focusing on transitions"},{"startTime":781.53,"endTime":785.655,"body":"is quite an important one."},{"startTime":785.655,"endTime":789.63,"body":"So how are patients able to"},{"startTime":785.655,"endTime":789.63,"body":"move between health states?"},{"startTime":789.63,"endTime":792.03,"body":"Also the incorporation"},{"startTime":789.63,"endTime":792.03,"body":"of time in the model"},{"startTime":792.03,"endTime":793.939,"body":"in terms of model cycle length"},{"startTime":793.939,"endTime":796.11,"body":"and also the time horizon."},{"startTime":796.11,"endTime":798.24,"body":"So how long are you evaluating the costs"},{"startTime":798.24,"endTime":799.56,"body":"and benefits over?"},{"startTime":799.56,"endTime":801.48,"body":"Usually, a lifetime horizon's used,"},{"startTime":801.48,"endTime":805.218,"body":"so that's up until patients are"},{"startTime":801.48,"endTime":805.218,"body":"roughly a hundred years old."},{"startTime":805.218,"endTime":808.77,"body":"But in some cases there's an"},{"startTime":805.218,"endTime":808.77,"body":"argument that shorter horizons"},{"startTime":808.77,"endTime":811.86,"body":"such as 10 or 20 years can be acceptable."},{"startTime":811.86,"endTime":813.93,"body":"Cases like this include"},{"startTime":811.86,"endTime":813.93,"body":"when you're looking"},{"startTime":813.93,"endTime":817.175,"body":"at a disease area which"},{"startTime":813.93,"endTime":817.175,"body":"is quite aggressive"},{"startTime":817.175,"endTime":820.83,"body":"in terms of progression"},{"startTime":817.175,"endTime":820.83,"body":"and the overall survival."},{"startTime":820.83,"endTime":822.21,"body":"So the number of patients alive"},{"startTime":822.21,"endTime":824.43,"body":"in a model decreases quite sharply."},{"startTime":824.43,"endTime":828.15,"body":"So there's no real benefit"},{"startTime":824.43,"endTime":828.15,"body":"to having a time horizon"},{"startTime":828.15,"endTime":829.83,"body":"of longer than 10 or 20 years"},{"startTime":829.83,"endTime":832.513,"body":"in particular from a"},{"startTime":829.83,"endTime":832.513,"body":"computational point of view,"},{"startTime":832.513,"endTime":837.09,"body":"you want Excel basically to"},{"startTime":832.513,"endTime":837.09,"body":"be able to run quite quickly"},{"startTime":837.09,"endTime":839.34,"body":"and you don't want to have"},{"startTime":837.09,"endTime":839.34,"body":"unnecessary data in there"},{"startTime":839.34,"endTime":841.09,"body":"if there's no justification for it."},{"startTime":842.28,"endTime":844.74,"body":"Get outputs from the model,"},{"startTime":842.28,"endTime":844.74,"body":"so we've got total costs"},{"startTime":844.74,"endTime":847.05,"body":"and total QALYs is the key ones here."},{"startTime":847.05,"endTime":848.88,"body":"So total costs being the cost accrued"},{"startTime":848.88,"endTime":850.77,"body":"over the defined time horizon."},{"startTime":850.77,"endTime":852.51,"body":"Likewise for total QALYs."},{"startTime":852.51,"endTime":855.518,"body":"So applying the utility"},{"startTime":852.51,"endTime":855.518,"body":"in each model cycle"},{"startTime":855.518,"endTime":858.45,"body":"to say, let's say we have yearly cycles"},{"startTime":858.45,"endTime":861.36,"body":"and be able to estimate"},{"startTime":858.45,"endTime":861.36,"body":"the life years accrued."},{"startTime":861.36,"endTime":863.96,"body":"So let's say it's 0.8 of a life year,"},{"startTime":863.96,"endTime":867.0,"body":"so patients are 0.8 of a"},{"startTime":863.96,"endTime":867.0,"body":"patient population are alive"},{"startTime":867.0,"endTime":869.675,"body":"in let's say the second model cycle."},{"startTime":869.675,"endTime":871.89,"body":"You can then apply utilities to that"},{"startTime":871.89,"endTime":873.72,"body":"based on the states that they're in"},{"startTime":873.72,"endTime":875.613,"body":"and estimate total QALY from there."},{"startTime":876.48,"endTime":879.87,"body":"The incremental cross effectiveness"},{"startTime":876.48,"endTime":879.87,"body":"ratio takes a difference"},{"startTime":879.87,"endTime":882.616,"body":"between the total costs of an"},{"startTime":879.87,"endTime":882.616,"body":"interventional and comparator"},{"startTime":882.616,"endTime":886.17,"body":"and divides that by the total QALYs"},{"startTime":886.17,"endTime":887.81,"body":"between an intervention and comparator"},{"startTime":887.81,"endTime":889.706,"body":"to get a cost of QALY ratio."},{"startTime":889.706,"endTime":892.571,"body":"And disease-specific"},{"startTime":889.706,"endTime":892.571,"body":"outcomes are also often used."},{"startTime":892.571,"endTime":895.2,"body":"These vary depending on the disease area"},{"startTime":895.2,"endTime":897.33,"body":"but it could be quite"},{"startTime":895.2,"endTime":897.33,"body":"important when you're looking"},{"startTime":897.33,"endTime":899.311,"body":"at specific discussions for an indication"},{"startTime":899.311,"endTime":902.613,"body":"and they can be very helpful"},{"startTime":899.311,"endTime":902.613,"body":"to a submission if collected."},{"startTime":903.72,"endTime":905.02,"body":"Next slide please, Hannah."},{"startTime":908.097,"endTime":912.06,"body":"So this slide talks about"},{"startTime":908.097,"endTime":912.06,"body":"the importance of accuracy"},{"startTime":912.06,"endTime":914.549,"body":"when reflecting a real world setting."},{"startTime":914.549,"endTime":918.69,"body":"So four key areas which"},{"startTime":914.549,"endTime":918.69,"body":"highlight the importance"},{"startTime":918.69,"endTime":920.88,"body":"of accuracy include being able"},{"startTime":920.88,"endTime":923.393,"body":"to capture clinically relevant endpoints."},{"startTime":923.393,"endTime":927.72,"body":"Examples here being that you"},{"startTime":923.393,"endTime":927.72,"body":"want to collect endpoints"},{"startTime":927.72,"endTime":930.9,"body":"which are most commonly used"},{"startTime":927.72,"endTime":930.9,"body":"in a clinical environment."},{"startTime":930.9,"endTime":932.34,"body":"You want those that are most relevant"},{"startTime":932.34,"endTime":935.758,"body":"to the clinical setting"},{"startTime":932.34,"endTime":935.758,"body":"and decision makers."},{"startTime":935.758,"endTime":938.91,"body":"It's also important to collect endpoints"},{"startTime":938.91,"endTime":940.677,"body":"which are common across clinical trials."},{"startTime":940.677,"endTime":942.42,"body":"So, those of competitors"},{"startTime":942.42,"endTime":944.43,"body":"to say if you have a single arm trial,"},{"startTime":944.43,"endTime":946.744,"body":"so where it's just your"},{"startTime":944.43,"endTime":946.744,"body":"treatment in the trial,"},{"startTime":946.744,"endTime":950.943,"body":"if you want to compare it"},{"startTime":946.744,"endTime":950.943,"body":"to a treatment that's used"},{"startTime":950.943,"endTime":952.38,"body":"in a different trial,"},{"startTime":952.38,"endTime":954.673,"body":"but which is the current,"},{"startTime":952.38,"endTime":954.673,"body":"say, standard of care,"},{"startTime":954.673,"endTime":957.43,"body":"so the current sort of"},{"startTime":954.673,"endTime":957.43,"body":"established best practise"},{"startTime":957.43,"endTime":958.696,"body":"in an indication,"},{"startTime":958.696,"endTime":962.88,"body":"having similar endpoints to"},{"startTime":958.696,"endTime":962.88,"body":"that trial would be important"},{"startTime":962.88,"endTime":964.48,"body":"to make comparisons of efficacy."},{"startTime":966.18,"endTime":968.76,"body":"The second point here"},{"startTime":966.18,"endTime":968.76,"body":"is patient transitions."},{"startTime":968.76,"endTime":970.53,"body":"So assumptions about the health states"},{"startTime":970.53,"endTime":971.883,"body":"that patients can move to."},{"startTime":973.14,"endTime":976.026,"body":"An example here would be"},{"startTime":973.14,"endTime":976.026,"body":"if you've got a patient"},{"startTime":976.026,"endTime":979.71,"body":"that's progressed, so"},{"startTime":976.026,"endTime":979.71,"body":"in an oncology setting,"},{"startTime":979.71,"endTime":983.19,"body":"a common approach is to have"},{"startTime":979.71,"endTime":983.19,"body":"progression-free health states."},{"startTime":983.19,"endTime":984.81,"body":"So when patients are receiving a treatment"},{"startTime":984.81,"endTime":986.972,"body":"but haven't progressed from the treatment."},{"startTime":986.972,"endTime":989.369,"body":"And when you have a progressed setting,"},{"startTime":989.369,"endTime":990.75,"body":"that's when patients have moved"},{"startTime":990.75,"endTime":993.72,"body":"from that progression-free"},{"startTime":990.75,"endTime":993.72,"body":"setting to a progressed state,"},{"startTime":993.72,"endTime":996.0,"body":"a much deteriorated health state."},{"startTime":996.0,"endTime":998.19,"body":"Patients then can't move from progressed"},{"startTime":998.19,"endTime":1000.905,"body":"to progression-free usually"},{"startTime":998.19,"endTime":1000.905,"body":"in most small structures"},{"startTime":1000.905,"endTime":1003.26,"body":"and it's about assessing the applicability"},{"startTime":1003.26,"endTime":1004.86,"body":"of that in a real world setting."},{"startTime":1006.2,"endTime":1009.44,"body":"Structural assumptions,"},{"startTime":1006.2,"endTime":1009.44,"body":"so incorrect assumptions"},{"startTime":1009.44,"endTime":1011.78,"body":"or structures that are"},{"startTime":1009.44,"endTime":1011.78,"body":"included in the type"},{"startTime":1011.78,"endTime":1014.952,"body":"of model selected, and the"},{"startTime":1011.78,"endTime":1014.952,"body":"flow, the sort of framework."},{"startTime":1014.952,"endTime":1019.155,"body":"Incorrect use of that can"},{"startTime":1014.952,"endTime":1019.155,"body":"add bias into model outputs"},{"startTime":1019.155,"endTime":1024.155,"body":"through incorrect assumptions"},{"startTime":1019.155,"endTime":1024.155,"body":"and also structural bias."},{"startTime":1024.457,"endTime":1028.211,"body":"Examples of this being if the transitions,"},{"startTime":1028.211,"endTime":1029.96,"body":"so going back to patient transitions,"},{"startTime":1029.96,"endTime":1030.793,"body":"if you've got a structure"},{"startTime":1030.793,"endTime":1032.87,"body":"which doesn't accurately"},{"startTime":1030.793,"endTime":1032.87,"body":"reflect the real world setting"},{"startTime":1032.87,"endTime":1035.81,"body":"and has incorrect patient"},{"startTime":1032.87,"endTime":1035.81,"body":"movements between states"},{"startTime":1035.81,"endTime":1038.748,"body":"for example, then it"},{"startTime":1035.81,"endTime":1038.748,"body":"will bias the outcomes"},{"startTime":1038.748,"endTime":1040.91,"body":"or it will likely bias the outcomes"},{"startTime":1040.91,"endTime":1043.43,"body":"because you're not accurately reflecting"},{"startTime":1043.43,"endTime":1046.1,"body":"what the real life setting looks like."},{"startTime":1046.1,"endTime":1048.551,"body":"And then accrual of costs"},{"startTime":1046.1,"endTime":1048.551,"body":"and health benefits."},{"startTime":1048.551,"endTime":1051.08,"body":"So this focuses on the timings"},{"startTime":1051.08,"endTime":1053.585,"body":"of accumulating the costs and the QALYs."},{"startTime":1053.585,"endTime":1058.31,"body":"An example, going back to"},{"startTime":1053.585,"endTime":1058.31,"body":"the example of a indication"},{"startTime":1058.31,"endTime":1060.247,"body":"where you've got very aggressive disease,"},{"startTime":1060.247,"endTime":1064.85,"body":"if you see a lot of changes in a clinical"},{"startTime":1064.85,"endTime":1066.68,"body":"or a disease pathway early on,"},{"startTime":1066.68,"endTime":1069.388,"body":"let's say within the first"},{"startTime":1066.68,"endTime":1069.388,"body":"couple of months of a model."},{"startTime":1069.388,"endTime":1072.2,"body":"However, if you are"},{"startTime":1069.388,"endTime":1072.2,"body":"using a year model cycle"},{"startTime":1072.2,"endTime":1073.67,"body":"so you're only assessing the costs"},{"startTime":1073.67,"endTime":1076.82,"body":"and benefits every year,"},{"startTime":1073.67,"endTime":1076.82,"body":"then you can lose quite a bit"},{"startTime":1076.82,"endTime":1079.069,"body":"of the intricacy that can occur early on"},{"startTime":1079.069,"endTime":1081.988,"body":"in that disease stage."},{"startTime":1081.988,"endTime":1086.39,"body":"And it's usually, or it is"},{"startTime":1081.988,"endTime":1086.39,"body":"recommended that you capture"},{"startTime":1086.39,"endTime":1088.136,"body":"as much of the sort of granularity"},{"startTime":1088.136,"endTime":1092.15,"body":"of the differences in"},{"startTime":1088.136,"endTime":1092.15,"body":"the disease early on."},{"startTime":1092.15,"endTime":1094.185,"body":"It's also important with model cycles"},{"startTime":1094.185,"endTime":1098.294,"body":"that they are a risk reasonable length"},{"startTime":1098.294,"endTime":1100.67,"body":"for say computational purposes as well."},{"startTime":1100.67,"endTime":1103.97,"body":"So for example, if you've"},{"startTime":1100.67,"endTime":1103.97,"body":"got a cycle length of a day,"},{"startTime":1103.97,"endTime":1107.27,"body":"then you are capturing"},{"startTime":1103.97,"endTime":1107.27,"body":"costs and QALYs each day"},{"startTime":1107.27,"endTime":1108.5,"body":"and if you've got a lifetime horizon,"},{"startTime":1108.5,"endTime":1111.41,"body":"that will be a very big"},{"startTime":1108.5,"endTime":1111.41,"body":"burden let's say on a model"},{"startTime":1111.41,"endTime":1113.663,"body":"and it could impact for"},{"startTime":1111.41,"endTime":1113.663,"body":"example, memory size."},{"startTime":1114.652,"endTime":1116.21,"body":"So it's important to"},{"startTime":1114.652,"endTime":1116.21,"body":"weigh your pros and cons"},{"startTime":1116.21,"endTime":1117.31,"body":"when approaching that."},{"startTime":1119.144,"endTime":1121.607,"body":"Thanks Hannah. Next slide."},{"startTime":1124.287,"endTime":1127.22,"body":"- Okay, so it's over to"},{"startTime":1124.287,"endTime":1127.22,"body":"me for the next section."},{"startTime":1127.22,"endTime":1129.68,"body":"So there are many different ways"},{"startTime":1129.68,"endTime":1132.29,"body":"that health economic"},{"startTime":1129.68,"endTime":1132.29,"body":"models can be classified"},{"startTime":1132.29,"endTime":1135.05,"body":"and often different"},{"startTime":1132.29,"endTime":1135.05,"body":"terminologies can be used"},{"startTime":1135.05,"endTime":1137.51,"body":"to describe the same model structure."},{"startTime":1137.51,"endTime":1140.21,"body":"So for example, some model"},{"startTime":1137.51,"endTime":1140.21,"body":"structures can be built"},{"startTime":1140.21,"endTime":1143.51,"body":"using either a cohort or"},{"startTime":1140.21,"endTime":1143.51,"body":"a patient-level approach,"},{"startTime":1143.51,"endTime":1146.03,"body":"which means that using the"},{"startTime":1143.51,"endTime":1146.03,"body":"term patient-level model"},{"startTime":1146.03,"endTime":1148.31,"body":"doesn't necessarily"},{"startTime":1146.03,"endTime":1148.31,"body":"describe its structure."},{"startTime":1148.31,"endTime":1150.333,"body":"Also, for example,"},{"startTime":1150.333,"endTime":1153.05,"body":"a state transition model"},{"startTime":1150.333,"endTime":1153.05,"body":"is an umbrella term"},{"startTime":1153.05,"endTime":1155.314,"body":"of which a Markov model is just one type"},{"startTime":1155.314,"endTime":1158.966,"body":"and a partition survival"},{"startTime":1155.314,"endTime":1158.966,"body":"model is also known"},{"startTime":1158.966,"endTime":1161.84,"body":"as an area under the curve model."},{"startTime":1161.84,"endTime":1164.504,"body":"So these kind of differences"},{"startTime":1164.504,"endTime":1165.337,"body":"and details can be confusing,"},{"startTime":1165.337,"endTime":1167.18,"body":"especially when you're new to the topic."},{"startTime":1167.18,"endTime":1168.787,"body":"And although we won't be able to cover all"},{"startTime":1168.787,"endTime":1172.04,"body":"of the possible types, we"},{"startTime":1168.787,"endTime":1172.04,"body":"just wanted to describe some"},{"startTime":1172.04,"endTime":1173.9,"body":"of the key model classifications"},{"startTime":1173.9,"endTime":1175.85,"body":"and structures during this webinar"},{"startTime":1175.85,"endTime":1179.483,"body":"to hopefully bring some"},{"startTime":1175.85,"endTime":1179.483,"body":"clarity to this topic."},{"startTime":1181.61,"endTime":1185.0,"body":"So firstly we'll describe the"},{"startTime":1181.61,"endTime":1185.0,"body":"difference between a cohort"},{"startTime":1185.0,"endTime":1187.55,"body":"and a patient-level approach."},{"startTime":1187.55,"endTime":1189.42,"body":"As I mentioned, some structures,"},{"startTime":1189.42,"endTime":1191.69,"body":"some model structures"},{"startTime":1189.42,"endTime":1191.69,"body":"do have the possibility"},{"startTime":1191.69,"endTime":1192.89,"body":"to take either approach,"},{"startTime":1192.89,"endTime":1195.74,"body":"so it's worth bearing"},{"startTime":1192.89,"endTime":1195.74,"body":"that in mind as well."},{"startTime":1195.74,"endTime":1197.901,"body":"So starting with cohort models,"},{"startTime":1197.901,"endTime":1202.13,"body":"this involves estimating"},{"startTime":1197.901,"endTime":1202.13,"body":"outcomes for a cohort,"},{"startTime":1202.13,"endTime":1204.68,"body":"so a whole average population,"},{"startTime":1204.68,"endTime":1207.979,"body":"without considering outcomes"},{"startTime":1204.68,"endTime":1207.979,"body":"for individual patients."},{"startTime":1207.979,"endTime":1209.81,"body":"And this is very common approach"},{"startTime":1209.81,"endTime":1212.87,"body":"in health economic evaluation."},{"startTime":1212.87,"endTime":1217.04,"body":"So a cohort model will take"},{"startTime":1212.87,"endTime":1217.04,"body":"the average population values"},{"startTime":1217.04,"endTime":1218.03,"body":"as their inputs."},{"startTime":1218.03,"endTime":1222.101,"body":"So for example the mean age"},{"startTime":1218.03,"endTime":1222.101,"body":"of the whole of the population"},{"startTime":1222.101,"endTime":1223.85,"body":"and this will be used"},{"startTime":1223.85,"endTime":1226.79,"body":"to estimate the total"},{"startTime":1223.85,"endTime":1226.79,"body":"population QALYs and costs"},{"startTime":1226.79,"endTime":1229.04,"body":"or the outcomes of the model."},{"startTime":1229.04,"endTime":1232.34,"body":"The main advantage is that"},{"startTime":1229.04,"endTime":1232.34,"body":"they are very transparent"},{"startTime":1232.34,"endTime":1235.16,"body":"and relatively straightforward"},{"startTime":1232.34,"endTime":1235.16,"body":"to develop, analyse"},{"startTime":1235.16,"endTime":1236.213,"body":"and to communicate."},{"startTime":1237.05,"endTime":1239.643,"body":"But a disadvantage is"},{"startTime":1237.05,"endTime":1239.643,"body":"that they don't reflect"},{"startTime":1239.643,"endTime":1242.121,"body":"the heterogeneity in a patient population,"},{"startTime":1242.121,"endTime":1245.704,"body":"which is the differences"},{"startTime":1242.121,"endTime":1245.704,"body":"between individual patients"},{"startTime":1245.704,"endTime":1248.084,"body":"because the average value"},{"startTime":1248.084,"endTime":1250.643,"body":"for the whole population is estimated."},{"startTime":1253.91,"endTime":1256.85,"body":"So in contrast, in a"},{"startTime":1253.91,"endTime":1256.85,"body":"patient-level approach,"},{"startTime":1256.85,"endTime":1259.641,"body":"the outcomes are modelled"},{"startTime":1256.85,"endTime":1259.641,"body":"for individual patients"},{"startTime":1259.641,"endTime":1262.46,"body":"and the results are based on the average"},{"startTime":1262.46,"endTime":1266.571,"body":"of a large sample of simulated patients."},{"startTime":1266.571,"endTime":1269.987,"body":"The advantage of the patient"},{"startTime":1266.571,"endTime":1269.987,"body":"level approach is that"},{"startTime":1269.987,"endTime":1273.348,"body":"it can capture the relationship"},{"startTime":1269.987,"endTime":1273.348,"body":"between the heterogeneity"},{"startTime":1273.348,"endTime":1276.86,"body":"in the patient characteristics"},{"startTime":1273.348,"endTime":1276.86,"body":"and the model outcomes,"},{"startTime":1276.86,"endTime":1280.325,"body":"but you do require individual patient data"},{"startTime":1280.325,"endTime":1284.3,"body":"in order to develop a patient-level model."},{"startTime":1284.3,"endTime":1288.023,"body":"Historically, access to this"},{"startTime":1284.3,"endTime":1288.023,"body":"kind of data was limited,"},{"startTime":1288.023,"endTime":1290.09,"body":"but the benefits"},{"startTime":1290.09,"endTime":1291.77,"body":"of the patient-level"},{"startTime":1290.09,"endTime":1291.77,"body":"models have really helped"},{"startTime":1291.77,"endTime":1294.593,"body":"to reduce these data access hurdles."},{"startTime":1295.983,"endTime":1298.07,"body":"Also, it is worth mentioning"},{"startTime":1298.07,"endTime":1301.721,"body":"that patient-level approaches"},{"startTime":1298.07,"endTime":1301.721,"body":"can be computationally heavy,"},{"startTime":1301.721,"endTime":1305.99,"body":"particularly when performing"},{"startTime":1301.721,"endTime":1305.99,"body":"partition probabilistic"},{"startTime":1305.99,"endTime":1307.702,"body":"sensitivity analysis,"},{"startTime":1307.702,"endTime":1311.18,"body":"which is when the parameters"},{"startTime":1307.702,"endTime":1311.18,"body":"in the model are varied"},{"startTime":1311.18,"endTime":1315.561,"body":"simultaneously to explore the"},{"startTime":1311.18,"endTime":1315.561,"body":"uncertainty of the parameters."},{"startTime":1315.561,"endTime":1320.561,"body":"So this combined with there"},{"startTime":1315.561,"endTime":1320.561,"body":"being one patient-level data run"},{"startTime":1321.583,"endTime":1324.89,"body":"through the model at a time"},{"startTime":1321.583,"endTime":1324.89,"body":"can cause longer runtimes"},{"startTime":1324.89,"endTime":1327.23,"body":"of these analyses, but"},{"startTime":1324.89,"endTime":1327.23,"body":"there are lots of way"},{"startTime":1327.23,"endTime":1331.493,"body":"to also make this process more efficient."},{"startTime":1334.49,"endTime":1337.04,"body":"So now that we've discussed"},{"startTime":1334.49,"endTime":1337.04,"body":"these two key approaches"},{"startTime":1337.04,"endTime":1339.05,"body":"to modelling, we're now going to move"},{"startTime":1339.05,"endTime":1341.562,"body":"into explaining our six model structures"},{"startTime":1341.562,"endTime":1343.659,"body":"that we've chosen to go through today"},{"startTime":1343.659,"endTime":1347.09,"body":"and I will hand over to Danny"},{"startTime":1347.09,"endTime":1349.883,"body":"to go through the first one"},{"startTime":1347.09,"endTime":1349.883,"body":"which are decision trees."},{"startTime":1351.53,"endTime":1352.507,"body":"- Thanks Hannah."},{"startTime":1353.498,"endTime":1355.523,"body":"So as Hannah mentioned, I'm"},{"startTime":1353.498,"endTime":1355.523,"body":"going to be starting off"},{"startTime":1355.523,"endTime":1357.147,"body":"with decision trees."},{"startTime":1357.147,"endTime":1360.2,"body":"So decision trees are very much a sort"},{"startTime":1360.2,"endTime":1364.602,"body":"of commonly used approach in"},{"startTime":1360.2,"endTime":1364.602,"body":"health economic modelling."},{"startTime":1364.602,"endTime":1366.17,"body":"I'd say decision trees"},{"startTime":1366.17,"endTime":1369.164,"body":"and Markov models are probably"},{"startTime":1366.17,"endTime":1369.164,"body":"the most commonly discussed"},{"startTime":1369.164,"endTime":1372.942,"body":"when discussing potential"},{"startTime":1369.164,"endTime":1372.942,"body":"modelling approaches."},{"startTime":1372.942,"endTime":1376.782,"body":"So to give a bit of an"},{"startTime":1372.942,"endTime":1376.782,"body":"overview of decision tree,"},{"startTime":1376.782,"endTime":1380.345,"body":"it's very much set up in"},{"startTime":1376.782,"endTime":1380.345,"body":"terms of different branches"},{"startTime":1380.345,"endTime":1383.103,"body":"modelling discrete events."},{"startTime":1383.103,"endTime":1385.4,"body":"So we've got a diagram here."},{"startTime":1385.4,"endTime":1386.666,"body":"So starting off,"},{"startTime":1386.666,"endTime":1389.09,"body":"the start of a decision tree discusses"},{"startTime":1389.09,"endTime":1391.984,"body":"whether a patient's"},{"startTime":1389.09,"endTime":1391.984,"body":"receiving prophylaxis or not."},{"startTime":1391.984,"endTime":1393.92,"body":"With a decision tree,"},{"startTime":1393.92,"endTime":1396.38,"body":"what then happens is"},{"startTime":1393.92,"endTime":1396.38,"body":"you reach a chance node"},{"startTime":1396.38,"endTime":1400.727,"body":"which allows you to go via a different"},{"startTime":1400.727,"endTime":1403.67,"body":"or compare two different options."},{"startTime":1403.67,"endTime":1406.91,"body":"So in this instance, you"},{"startTime":1403.67,"endTime":1406.91,"body":"can go up to no event,"},{"startTime":1406.91,"endTime":1408.981,"body":"so whether you've received an event."},{"startTime":1408.981,"endTime":1412.04,"body":"Or you can go to the other sort of branch"},{"startTime":1412.04,"endTime":1415.463,"body":"which will go to venous thrombus embolism."},{"startTime":1416.325,"endTime":1417.68,"body":"With decision trees,"},{"startTime":1417.68,"endTime":1419.78,"body":"what you then do is you"},{"startTime":1417.68,"endTime":1419.78,"body":"then move through a series"},{"startTime":1419.78,"endTime":1423.302,"body":"of events which classify"},{"startTime":1419.78,"endTime":1423.302,"body":"into different categories."},{"startTime":1423.302,"endTime":1425.75,"body":"The idea being that you continue"},{"startTime":1425.75,"endTime":1427.019,"body":"along the different branches"},{"startTime":1427.019,"endTime":1429.95,"body":"relating to what status you are in"},{"startTime":1429.95,"endTime":1432.045,"body":"until you reach the end of the branch"},{"startTime":1432.045,"endTime":1433.763,"body":"known as the terminal node."},{"startTime":1434.66,"endTime":1435.68,"body":"So these will be denoted"},{"startTime":1435.68,"endTime":1438.729,"body":"by sort of the triangle"},{"startTime":1435.68,"endTime":1438.729,"body":"symbols at the end,"},{"startTime":1438.729,"endTime":1443.15,"body":"at which point you'll"},{"startTime":1438.729,"endTime":1443.15,"body":"accrue a cost and QALY."},{"startTime":1443.15,"endTime":1444.903,"body":"So QALY represents a health benefit"},{"startTime":1444.903,"endTime":1447.35,"body":"for that specific branch."},{"startTime":1447.35,"endTime":1449.42,"body":"There are also probabilities assigned"},{"startTime":1449.42,"endTime":1451.16,"body":"to each specific branch."},{"startTime":1451.16,"endTime":1453.59,"body":"So the probability of"},{"startTime":1451.16,"endTime":1453.59,"body":"a patient for example"},{"startTime":1453.59,"endTime":1456.501,"body":"going to either no event or to VTE,"},{"startTime":1456.501,"endTime":1460.298,"body":"the probabilities of some"},{"startTime":1456.501,"endTime":1460.298,"body":"to one in that instance."},{"startTime":1460.298,"endTime":1462.95,"body":"Showing the different"},{"startTime":1460.298,"endTime":1462.95,"body":"options that you have,"},{"startTime":1462.95,"endTime":1466.31,"body":"you can then weight the"},{"startTime":1462.95,"endTime":1466.31,"body":"QALY and cost outcome"},{"startTime":1466.31,"endTime":1468.32,"body":"at the terminal nodes by the probability"},{"startTime":1468.32,"endTime":1472.78,"body":"of a patient ending up"},{"startTime":1468.32,"endTime":1472.78,"body":"at that specific node."},{"startTime":1472.78,"endTime":1475.88,"body":"And then what you do is you"},{"startTime":1472.78,"endTime":1475.88,"body":"take the weighted average"},{"startTime":1475.88,"endTime":1478.502,"body":"across each of the"},{"startTime":1475.88,"endTime":1478.502,"body":"specific terminal nodes,"},{"startTime":1478.502,"endTime":1481.498,"body":"so weighted by the probability"},{"startTime":1478.502,"endTime":1481.498,"body":"of patients getting"},{"startTime":1481.498,"endTime":1483.35,"body":"to that specific node"},{"startTime":1483.35,"endTime":1485.828,"body":"and then you are able"},{"startTime":1483.35,"endTime":1485.828,"body":"to estimate an estimated"},{"startTime":1485.828,"endTime":1489.65,"body":"or sorry, an expected cost"},{"startTime":1485.828,"endTime":1489.65,"body":"and an expected outcome."},{"startTime":1489.65,"endTime":1490.853,"body":"So an expected QALY."},{"startTime":1492.56,"endTime":1496.25,"body":"For specific strengths and"},{"startTime":1492.56,"endTime":1496.25,"body":"limitations to decision trees."},{"startTime":1496.25,"endTime":1499.07,"body":"So if we start with a"},{"startTime":1496.25,"endTime":1499.07,"body":"strengths, they provide a clear"},{"startTime":1499.07,"endTime":1500.78,"body":"and transparent representation"},{"startTime":1500.78,"endTime":1503.06,"body":"of the decision making process."},{"startTime":1503.06,"endTime":1506.18,"body":"So in instances such as"},{"startTime":1503.06,"endTime":1506.18,"body":"the one that we have here"},{"startTime":1506.18,"endTime":1509.69,"body":"with the diagram, it's clear"},{"startTime":1506.18,"endTime":1509.69,"body":"to map out the specific options"},{"startTime":1509.69,"endTime":1511.4,"body":"that a patient has at each point"},{"startTime":1511.4,"endTime":1512.63,"body":"and you're able to clearly"},{"startTime":1512.63,"endTime":1515.066,"body":"and transparently move along the model"},{"startTime":1515.066,"endTime":1519.26,"body":"to reach the state at"},{"startTime":1515.066,"endTime":1519.26,"body":"which the patient's at."},{"startTime":1519.26,"endTime":1520.902,"body":"They are highly adaptable"},{"startTime":1520.902,"endTime":1524.48,"body":"and can model a wide range"},{"startTime":1520.902,"endTime":1524.48,"body":"of healthcare scenarios."},{"startTime":1524.48,"endTime":1526.58,"body":"So these can be commonplace"},{"startTime":1524.48,"endTime":1526.58,"body":"across different"},{"startTime":1526.58,"endTime":1530.24,"body":"as say indications, not"},{"startTime":1526.58,"endTime":1530.24,"body":"even related to sort"},{"startTime":1530.24,"endTime":1533.24,"body":"of healthcare-specific"},{"startTime":1530.24,"endTime":1533.24,"body":"cross factors modelling."},{"startTime":1533.24,"endTime":1535.04,"body":"They are very commonly used."},{"startTime":1535.04,"endTime":1537.474,"body":"They're a logical sort of"},{"startTime":1535.04,"endTime":1537.474,"body":"flow diagram approaches"},{"startTime":1537.474,"endTime":1540.83,"body":"and they're very, as I said, adaptable"},{"startTime":1540.83,"endTime":1542.18,"body":"across different scenarios."},{"startTime":1543.32,"endTime":1546.205,"body":"With decision trees there"},{"startTime":1543.32,"endTime":1546.205,"body":"are some key limitations."},{"startTime":1546.205,"endTime":1548.905,"body":"So the first one being"},{"startTime":1546.205,"endTime":1548.905,"body":"that continuous variables"},{"startTime":1548.905,"endTime":1552.38,"body":"such as health states with continuous sort"},{"startTime":1552.38,"endTime":1556.072,"body":"of parameters being analysed"},{"startTime":1552.38,"endTime":1556.072,"body":"are difficult to model."},{"startTime":1556.072,"endTime":1560.764,"body":"So for example, if we have"},{"startTime":1556.072,"endTime":1560.764,"body":"a sort of chronic disease"},{"startTime":1560.764,"endTime":1564.525,"body":"that we're looking at patients that occur"},{"startTime":1564.525,"endTime":1568.044,"body":"or reoccur different sort"},{"startTime":1564.525,"endTime":1568.044,"body":"of clinical outcomes,"},{"startTime":1568.044,"endTime":1571.985,"body":"on a model cycle basis,"},{"startTime":1568.044,"endTime":1571.985,"body":"that'd be quite difficult"},{"startTime":1571.985,"endTime":1575.26,"body":"because the decisions tree"},{"startTime":1571.985,"endTime":1575.26,"body":"only specifically looks"},{"startTime":1575.26,"endTime":1577.7,"body":"at discrete events that we're looking at."},{"startTime":1577.7,"endTime":1580.1,"body":"So modelling long term is difficult."},{"startTime":1580.1,"endTime":1581.48,"body":"Another limitation is"},{"startTime":1581.48,"endTime":1584.96,"body":"that when the decision"},{"startTime":1581.48,"endTime":1584.96,"body":"tree becomes quite complex,"},{"startTime":1584.96,"endTime":1586.4,"body":"so there's specific, there's loads"},{"startTime":1586.4,"endTime":1588.83,"body":"of different branches that are required."},{"startTime":1588.83,"endTime":1590.69,"body":"So let's say if we expanded this three"},{"startTime":1590.69,"endTime":1593.3,"body":"or four times over, it"},{"startTime":1590.69,"endTime":1593.3,"body":"can be quite complicated"},{"startTime":1593.3,"endTime":1595.55,"body":"and complex to look at."},{"startTime":1595.55,"endTime":1599.69,"body":"In which case models such as"},{"startTime":1595.55,"endTime":1599.69,"body":"a Markov model might be used"},{"startTime":1599.69,"endTime":1602.69,"body":"where you're able to identify"},{"startTime":1599.69,"endTime":1602.69,"body":"a couple of health states"},{"startTime":1602.69,"endTime":1604.331,"body":"and then you are able to define them"},{"startTime":1604.331,"endTime":1607.22,"body":"and cover sort of each"},{"startTime":1607.22,"endTime":1610.12,"body":"of the different potential"},{"startTime":1607.22,"endTime":1610.12,"body":"events that a patient may have."},{"startTime":1611.06,"endTime":1613.722,"body":"So for the next model structure"},{"startTime":1611.06,"endTime":1613.722,"body":"that we'll be discussing,"},{"startTime":1613.722,"endTime":1615.677,"body":"Hannah's going to move into Markov models."},{"startTime":1615.677,"endTime":1619.25,"body":"And Markov models, like I said"},{"startTime":1615.677,"endTime":1619.25,"body":"with decision trees, are two"},{"startTime":1619.25,"endTime":1621.65,"body":"of the probably most common"},{"startTime":1619.25,"endTime":1621.65,"body":"approaches discussed."},{"startTime":1621.65,"endTime":1623.48,"body":"I'd probably say Markov"},{"startTime":1621.65,"endTime":1623.48,"body":"models slightly more"},{"startTime":1623.48,"endTime":1625.083,"body":"than decision trees,"},{"startTime":1625.083,"endTime":1628.823,"body":"but I'll allow Hannah to"},{"startTime":1625.083,"endTime":1628.823,"body":"go through and discuss it."},{"startTime":1630.11,"endTime":1631.981,"body":"- Yeah, that's great, thank you."},{"startTime":1631.981,"endTime":1635.75,"body":"So yeah, as Danny said,"},{"startTime":1631.981,"endTime":1635.75,"body":"you're probably familiar"},{"startTime":1635.75,"endTime":1639.23,"body":"with Markov models as"},{"startTime":1635.75,"endTime":1639.23,"body":"they are quite common."},{"startTime":1639.23,"endTime":1642.77,"body":"And just to mention that"},{"startTime":1639.23,"endTime":1642.77,"body":"they can take either a cohort"},{"startTime":1642.77,"endTime":1644.843,"body":"or a patient-level approach."},{"startTime":1644.843,"endTime":1648.11,"body":"And when a patient-level approach is taken"},{"startTime":1648.11,"endTime":1650.18,"body":"for a Markov model, this"},{"startTime":1648.11,"endTime":1650.18,"body":"can sometimes be referred to"},{"startTime":1650.18,"endTime":1651.847,"body":"as a microsimulation."},{"startTime":1656.204,"endTime":1658.73,"body":"Okay, so Markov models use health states"},{"startTime":1658.73,"endTime":1661.49,"body":"to represent all possible"},{"startTime":1658.73,"endTime":1661.49,"body":"mutually exclusive"},{"startTime":1661.49,"endTime":1664.46,"body":"and exhaustive disease health states."},{"startTime":1664.46,"endTime":1665.93,"body":"Each health state is associated"},{"startTime":1665.93,"endTime":1668.21,"body":"with a distinct cost and health outcome,"},{"startTime":1668.21,"endTime":1669.533,"body":"for example, utility."},{"startTime":1670.43,"endTime":1671.6,"body":"The time is considered"},{"startTime":1671.6,"endTime":1674.09,"body":"in discrete time periods called cycles"},{"startTime":1674.09,"endTime":1676.19,"body":"and the movements from one health state"},{"startTime":1676.19,"endTime":1679.28,"body":"to another are represented"},{"startTime":1676.19,"endTime":1679.28,"body":"as transition probabilities"},{"startTime":1679.28,"endTime":1682.31,"body":"which occur at the start"},{"startTime":1679.28,"endTime":1682.31,"body":"and the end of each cycle."},{"startTime":1682.31,"endTime":1685.01,"body":"So if we look at the example"},{"startTime":1682.31,"endTime":1685.01,"body":"on the right hand side,"},{"startTime":1685.01,"endTime":1689.72,"body":"this is an example of a simple"},{"startTime":1685.01,"endTime":1689.72,"body":"three state Markov model."},{"startTime":1689.72,"endTime":1692.27,"body":"In this example we have the blue circles"},{"startTime":1692.27,"endTime":1694.61,"body":"representing the health"},{"startTime":1692.27,"endTime":1694.61,"body":"states and the three"},{"startTime":1694.61,"endTime":1696.68,"body":"that we have here are progression-free,"},{"startTime":1696.68,"endTime":1698.75,"body":"progressed disease and dead."},{"startTime":1698.75,"endTime":1702.95,"body":"And this is a very common"},{"startTime":1698.75,"endTime":1702.95,"body":"Markov model used in modelling"},{"startTime":1702.95,"endTime":1704.96,"body":"of oncology and cancer treatments"},{"startTime":1704.96,"endTime":1709.49,"body":"where patients can be categorised"},{"startTime":1704.96,"endTime":1709.49,"body":"into these health states."},{"startTime":1709.49,"endTime":1711.421,"body":"And the arrows on the diagram represent"},{"startTime":1711.421,"endTime":1714.47,"body":"where the patient movement can happen"},{"startTime":1714.47,"endTime":1716.001,"body":"from one health state to another"},{"startTime":1716.001,"endTime":1718.94,"body":"and the probability of"},{"startTime":1716.001,"endTime":1718.94,"body":"the movement will be based"},{"startTime":1718.94,"endTime":1721.31,"body":"on the transition probabilities."},{"startTime":1721.31,"endTime":1724.49,"body":"You can also see that some"},{"startTime":1721.31,"endTime":1724.49,"body":"of the arrows go around"},{"startTime":1724.49,"endTime":1727.64,"body":"into themselves and this"},{"startTime":1724.49,"endTime":1727.64,"body":"represents the probability"},{"startTime":1727.64,"endTime":1730.97,"body":"of the patients remaining"},{"startTime":1727.64,"endTime":1730.97,"body":"within that health state"},{"startTime":1730.97,"endTime":1731.873,"body":"in that cycle."},{"startTime":1733.55,"endTime":1736.13,"body":"So, as I mentioned, the"},{"startTime":1733.55,"endTime":1736.13,"body":"probability of transition"},{"startTime":1736.13,"endTime":1741.13,"body":"to any particular state depends"},{"startTime":1736.13,"endTime":1741.13,"body":"solely on the current state"},{"startTime":1741.17,"endTime":1742.37,"body":"and the time elapsed."},{"startTime":1742.37,"endTime":1744.38,"body":"It doesn't depend on"},{"startTime":1742.37,"endTime":1744.38,"body":"the sequence of events"},{"startTime":1744.38,"endTime":1745.7,"body":"that preceded it."},{"startTime":1745.7,"endTime":1747.23,"body":"And this property is often referred to"},{"startTime":1747.23,"endTime":1748.52,"body":"as the memoryless property"},{"startTime":1748.52,"endTime":1750.14,"body":"or the Markovian assumption,"},{"startTime":1750.14,"endTime":1753.02,"body":"which is a very important"},{"startTime":1750.14,"endTime":1753.02,"body":"underlying assumption"},{"startTime":1753.02,"endTime":1754.163,"body":"of Markov models."},{"startTime":1755.57,"endTime":1758.87,"body":"They do typically use a cohort"},{"startTime":1755.57,"endTime":1758.87,"body":"approach where the costs"},{"startTime":1758.87,"endTime":1760.19,"body":"and health outcomes are aggregated"},{"startTime":1760.19,"endTime":1762.14,"body":"over the model time horizon"},{"startTime":1762.14,"endTime":1764.48,"body":"to provide a summary of"},{"startTime":1762.14,"endTime":1764.48,"body":"the cohort experience"},{"startTime":1764.48,"endTime":1766.37,"body":"and then the outcomes of this are compared"},{"startTime":1766.37,"endTime":1768.71,"body":"to those of another cohort"},{"startTime":1768.71,"endTime":1770.66,"body":"where you are receiving"},{"startTime":1768.71,"endTime":1770.66,"body":"a different intervention"},{"startTime":1770.66,"endTime":1771.953,"body":"to make that comparison."},{"startTime":1773.15,"endTime":1777.743,"body":"But as I mentioned, it is"},{"startTime":1773.15,"endTime":1777.743,"body":"possible to run one patient's data"},{"startTime":1777.743,"endTime":1779.444,"body":"through the model at a time"},{"startTime":1779.444,"endTime":1782.153,"body":"to achieve a patient-level"},{"startTime":1779.444,"endTime":1782.153,"body":"approach as well."},{"startTime":1783.414,"endTime":1786.83,"body":"And so the strengths"},{"startTime":1783.414,"endTime":1786.83,"body":"are that it's intuitive"},{"startTime":1786.83,"endTime":1789.498,"body":"and very widely used in"},{"startTime":1786.83,"endTime":1789.498,"body":"health economic evaluations."},{"startTime":1789.498,"endTime":1791.6,"body":"It's simple to estimate the costs"},{"startTime":1791.6,"endTime":1793.834,"body":"and health benefits on a per cycle basis."},{"startTime":1793.834,"endTime":1797.259,"body":"So for example, if you"},{"startTime":1793.834,"endTime":1797.259,"body":"have a monthly cycle,"},{"startTime":1797.259,"endTime":1800.278,"body":"you would need to source"},{"startTime":1797.259,"endTime":1800.278,"body":"the costs and utilities"},{"startTime":1800.278,"endTime":1804.32,"body":"and these can be applied"},{"startTime":1800.278,"endTime":1804.32,"body":"on a monthly basis"},{"startTime":1804.32,"endTime":1805.42,"body":"throughout your model."},{"startTime":1806.48,"endTime":1808.731,"body":"There are some limitations."},{"startTime":1808.731,"endTime":1812.39,"body":"The memoryless property that"},{"startTime":1808.731,"endTime":1812.39,"body":"I mentioned is very important."},{"startTime":1812.39,"endTime":1814.73,"body":"This means that the"},{"startTime":1812.39,"endTime":1814.73,"body":"transitions are independent"},{"startTime":1814.73,"endTime":1816.851,"body":"of the previous cycles"},{"startTime":1816.851,"endTime":1820.37,"body":"and the probabilities"},{"startTime":1816.851,"endTime":1820.37,"body":"of moving from one state"},{"startTime":1820.37,"endTime":1823.853,"body":"to another are therefore"},{"startTime":1820.37,"endTime":1823.853,"body":"assumed to remain constant."},{"startTime":1824.69,"endTime":1827.36,"body":"So there is a way to"},{"startTime":1824.69,"endTime":1827.36,"body":"overcome this limitation"},{"startTime":1827.36,"endTime":1829.67,"body":"and this is to, one of the ways to do this"},{"startTime":1829.67,"endTime":1832.292,"body":"is to add tunnel states into the model."},{"startTime":1832.292,"endTime":1834.292,"body":"And a lot of the time,"},{"startTime":1834.292,"endTime":1837.11,"body":"model developers will"},{"startTime":1834.292,"endTime":1837.11,"body":"relax these assumptions"},{"startTime":1837.11,"endTime":1839.78,"body":"by making adjust these"},{"startTime":1837.11,"endTime":1839.78,"body":"kinds of adjustments,"},{"startTime":1839.78,"endTime":1842.149,"body":"but it can quickly become cumbersome"},{"startTime":1842.149,"endTime":1845.72,"body":"and this will be discussed"},{"startTime":1842.149,"endTime":1845.72,"body":"further in our next example"},{"startTime":1845.72,"endTime":1849.383,"body":"which Daniel will explain,"},{"startTime":1845.72,"endTime":1849.383,"body":"which are semi-Markov models."},{"startTime":1851.06,"endTime":1852.54,"body":"- Thanks Hannah."},{"startTime":1852.54,"endTime":1857.027,"body":"So as Hannah mentioned,"},{"startTime":1852.54,"endTime":1857.027,"body":"semi-Markov models are an extension"},{"startTime":1857.027,"endTime":1858.89,"body":"to Markov models"},{"startTime":1858.89,"endTime":1863.12,"body":"where they specifically"},{"startTime":1858.89,"endTime":1863.12,"body":"look at the benefits"},{"startTime":1863.12,"endTime":1864.05,"body":"or the look at ways"},{"startTime":1864.05,"endTime":1866.25,"body":"which can overcome that"},{"startTime":1864.05,"endTime":1866.25,"body":"memoryless property."},{"startTime":1867.17,"endTime":1869.543,"body":"So please going to move"},{"startTime":1867.17,"endTime":1869.543,"body":"on to the next slide."},{"startTime":1871.04,"endTime":1874.94,"body":"So a semi-Markov model"},{"startTime":1871.04,"endTime":1874.94,"body":"is still built with sort"},{"startTime":1874.94,"endTime":1877.28,"body":"of a similar structure to a Markov model."},{"startTime":1877.28,"endTime":1880.271,"body":"So there's a diagram on"},{"startTime":1877.28,"endTime":1880.271,"body":"the right hand side here."},{"startTime":1880.271,"endTime":1882.65,"body":"For people who listen on audio,"},{"startTime":1882.65,"endTime":1885.391,"body":"there's three health states"},{"startTime":1882.65,"endTime":1885.391,"body":"or key health states here."},{"startTime":1885.391,"endTime":1890.391,"body":"So those are well, mild"},{"startTime":1885.391,"endTime":1890.391,"body":"symptoms or severe symptoms."},{"startTime":1891.08,"endTime":1895.88,"body":"However, as mentioned with a Markov model,"},{"startTime":1895.88,"endTime":1899.18,"body":"it does not acknowledge"},{"startTime":1895.88,"endTime":1899.18,"body":"the previous states"},{"startTime":1899.18,"endTime":1901.97,"body":"or the sort of history of movement"},{"startTime":1901.97,"endTime":1903.99,"body":"or transition between different states"},{"startTime":1903.99,"endTime":1906.08,"body":"that a patient could have."},{"startTime":1906.08,"endTime":1909.38,"body":"In some cases there are benefits"},{"startTime":1909.38,"endTime":1910.729,"body":"of incorporating that"},{"startTime":1910.729,"endTime":1914.45,"body":"and that's when a semi-Markov"},{"startTime":1910.729,"endTime":1914.45,"body":"model could be used."},{"startTime":1914.45,"endTime":1916.91,"body":"So an example of how"},{"startTime":1914.45,"endTime":1916.91,"body":"this can be incorporated"},{"startTime":1916.91,"endTime":1918.311,"body":"is through tunnel states."},{"startTime":1918.311,"endTime":1922.225,"body":"So as we can see here from"},{"startTime":1918.311,"endTime":1922.225,"body":"the mild symptom state,"},{"startTime":1922.225,"endTime":1926.144,"body":"there's the option of a"},{"startTime":1922.225,"endTime":1926.144,"body":"patient to move into one"},{"startTime":1926.144,"endTime":1928.73,"body":"of two tunnel states"},{"startTime":1926.144,"endTime":1928.73,"body":"which is called recovery."},{"startTime":1928.73,"endTime":1931.76,"body":"So M1 on the recovery is called mild"},{"startTime":1931.76,"endTime":1934.4,"body":"and then there's also"},{"startTime":1931.76,"endTime":1934.4,"body":"a second tunnel state"},{"startTime":1934.4,"endTime":1935.69,"body":"denoted as M2,"},{"startTime":1935.69,"endTime":1940.07,"body":"which is the second sort of"},{"startTime":1935.69,"endTime":1940.07,"body":"stage of mildness I'll say,"},{"startTime":1940.07,"endTime":1942.74,"body":"or slightly improved treatment outcome."},{"startTime":1942.74,"endTime":1947.12,"body":"In that recovery state what we"},{"startTime":1942.74,"endTime":1947.12,"body":"are able to do is we're able"},{"startTime":1947.12,"endTime":1950.9,"body":"to say, okay, so if we are"},{"startTime":1947.12,"endTime":1950.9,"body":"in the mild symptom state"},{"startTime":1950.9,"endTime":1953.12,"body":"and we want to show the"},{"startTime":1950.9,"endTime":1953.12,"body":"different sort of stages"},{"startTime":1953.12,"endTime":1956.211,"body":"of recovery, rather than just"},{"startTime":1953.12,"endTime":1956.211,"body":"having one way go from mild"},{"startTime":1956.211,"endTime":1959.09,"body":"say to well or what mild to severe,"},{"startTime":1959.09,"endTime":1960.26,"body":"what you're able to do is you're able"},{"startTime":1960.26,"endTime":1962.51,"body":"to assign these tunnel states,"},{"startTime":1962.51,"endTime":1964.76,"body":"assign costs and QALYs to them,"},{"startTime":1964.76,"endTime":1966.47,"body":"which denote the different sort of levels"},{"startTime":1966.47,"endTime":1968.15,"body":"of recovery you could have"},{"startTime":1968.15,"endTime":1971.548,"body":"and you're able to model"},{"startTime":1968.15,"endTime":1971.548,"body":"these in model cycles."},{"startTime":1971.548,"endTime":1975.08,"body":"So our resource requirements here,"},{"startTime":1975.08,"endTime":1977.65,"body":"so in a Markov model we have"},{"startTime":1975.08,"endTime":1977.65,"body":"transition probabilities."},{"startTime":1977.65,"endTime":1980.321,"body":"So the probability each"},{"startTime":1977.65,"endTime":1980.321,"body":"cycle of a patient moving"},{"startTime":1980.321,"endTime":1982.49,"body":"between health states."},{"startTime":1982.49,"endTime":1984.62,"body":"Having multiple tunnel states would"},{"startTime":1984.62,"endTime":1988.19,"body":"require transition"},{"startTime":1984.62,"endTime":1988.19,"body":"probabilities to be estimated"},{"startTime":1988.19,"endTime":1990.62,"body":"for movement into those states."},{"startTime":1990.62,"endTime":1993.899,"body":"In this diagram we have six"},{"startTime":1990.62,"endTime":1993.899,"body":"tunnel states, for example."},{"startTime":1993.899,"endTime":1997.25,"body":"So suddenly going from say three states"},{"startTime":1997.25,"endTime":1998.48,"body":"of well, mild symptoms,"},{"startTime":1998.48,"endTime":2001.12,"body":"and severe symptoms to"},{"startTime":1998.48,"endTime":2001.12,"body":"then having nine states,"},{"startTime":2001.12,"endTime":2004.09,"body":"six of which are"},{"startTime":2001.12,"endTime":2004.09,"body":"recovery-based tunnel states."},{"startTime":2004.09,"endTime":2005.386,"body":"So you're very much going,"},{"startTime":2005.386,"endTime":2008.17,"body":"you are adding sort of data requirements"},{"startTime":2008.17,"endTime":2013.12,"body":"for the transition probabilities"},{"startTime":2008.17,"endTime":2013.12,"body":"at the not expense,"},{"startTime":2013.12,"endTime":2016.7,"body":"but for the benefit of"},{"startTime":2013.12,"endTime":2016.7,"body":"having additional granularity"},{"startTime":2016.7,"endTime":2018.403,"body":"and detail in the model."},{"startTime":2020.2,"endTime":2022.33,"body":"So there's several strengths here."},{"startTime":2022.33,"endTime":2023.38,"body":"So there's flexibility"},{"startTime":2023.38,"endTime":2025.42,"body":"in representing complex health states."},{"startTime":2025.42,"endTime":2026.62,"body":"So in some cases,"},{"startTime":2026.62,"endTime":2029.68,"body":"health states may have a lot"},{"startTime":2026.62,"endTime":2029.68,"body":"of variation within them."},{"startTime":2029.68,"endTime":2031.99,"body":"So there might be a lot of"},{"startTime":2029.68,"endTime":2031.99,"body":"heterogeneity of disease"},{"startTime":2031.99,"endTime":2033.25,"body":"within the health states"},{"startTime":2033.25,"endTime":2035.642,"body":"that might be difficult to represent."},{"startTime":2035.642,"endTime":2037.9,"body":"When you only have a"},{"startTime":2035.642,"endTime":2037.9,"body":"specific health state,"},{"startTime":2037.9,"endTime":2040.57,"body":"you're looking to average"},{"startTime":2037.9,"endTime":2040.57,"body":"the values within that."},{"startTime":2040.57,"endTime":2042.61,"body":"So you'll be taking the"},{"startTime":2040.57,"endTime":2042.61,"body":"average of the different levels"},{"startTime":2042.61,"endTime":2044.92,"body":"of severity that could"},{"startTime":2042.61,"endTime":2044.92,"body":"occur within a state."},{"startTime":2044.92,"endTime":2046.78,"body":"If we have different tunnel"},{"startTime":2044.92,"endTime":2046.78,"body":"states, then what you're able"},{"startTime":2046.78,"endTime":2049.33,"body":"to do is you're able to"},{"startTime":2046.78,"endTime":2049.33,"body":"increase the granularity,"},{"startTime":2049.33,"endTime":2052.06,"body":"increase the detail, which allows"},{"startTime":2052.06,"endTime":2055.507,"body":"for a more specific representation"},{"startTime":2052.06,"endTime":2055.507,"body":"of the real world setting"},{"startTime":2055.507,"endTime":2059.413,"body":"and a more specific representation"},{"startTime":2055.507,"endTime":2059.413,"body":"of the disease pathway."},{"startTime":2061.06,"endTime":2064.017,"body":"It allows accurate representation"},{"startTime":2061.06,"endTime":2064.017,"body":"of time-dependent events."},{"startTime":2064.017,"endTime":2067.72,"body":"So you can incorporate events"},{"startTime":2064.017,"endTime":2067.72,"body":"which are say conditional"},{"startTime":2067.72,"endTime":2070.658,"body":"on a patient achieving a"},{"startTime":2067.72,"endTime":2070.658,"body":"certain level of health."},{"startTime":2070.658,"endTime":2075.38,"body":"So in these cases, for example"},{"startTime":2070.658,"endTime":2075.38,"body":"in the recovery M2 state,"},{"startTime":2075.38,"endTime":2079.617,"body":"you can incorporate an event"},{"startTime":2075.38,"endTime":2079.617,"body":"in either the recovery M1"},{"startTime":2079.617,"endTime":2081.7,"body":"or recovery M2 and you're able to reflect"},{"startTime":2081.7,"endTime":2084.403,"body":"how a patient would respond"},{"startTime":2081.7,"endTime":2084.403,"body":"to that specific event."},{"startTime":2085.36,"endTime":2087.4,"body":"It might be difficult to"},{"startTime":2085.36,"endTime":2087.4,"body":"incorporate these events"},{"startTime":2087.4,"endTime":2089.565,"body":"if you only have say a three state model,"},{"startTime":2089.565,"endTime":2092.14,"body":"because of these sort"},{"startTime":2089.565,"endTime":2092.14,"body":"of, as I said, variation"},{"startTime":2092.14,"endTime":2094.368,"body":"and heterogeneity in"},{"startTime":2092.14,"endTime":2094.368,"body":"the patient population."},{"startTime":2094.368,"endTime":2097.63,"body":"These also build into longer"},{"startTime":2094.368,"endTime":2097.63,"body":"term predictive power."},{"startTime":2097.63,"endTime":2099.22,"body":"So there are more, you'd assume,"},{"startTime":2099.22,"endTime":2100.87,"body":"a more accurate representation"},{"startTime":2100.87,"endTime":2102.94,"body":"of the disease pathway"},{"startTime":2100.87,"endTime":2102.94,"body":"in real world setting"},{"startTime":2102.94,"endTime":2105.082,"body":"which adds to the robustness of a model"},{"startTime":2105.082,"endTime":2108.13,"body":"and adds I'd say to the argument,"},{"startTime":2108.13,"endTime":2111.19,"body":"if you are approaching"},{"startTime":2108.13,"endTime":2111.19,"body":"sort of a HTA submission,"},{"startTime":2111.19,"endTime":2112.023,"body":"for example."},{"startTime":2112.96,"endTime":2115.36,"body":"The limitations here, the amount"},{"startTime":2115.36,"endTime":2117.661,"body":"of health states you have"},{"startTime":2115.36,"endTime":2117.661,"body":"adds a level of complexity"},{"startTime":2117.661,"endTime":2120.64,"body":"as opposed to a three state"},{"startTime":2117.661,"endTime":2120.64,"body":"model, which would be seen"},{"startTime":2120.64,"endTime":2122.19,"body":"as more sort of simple version."},{"startTime":2123.04,"endTime":2125.11,"body":"The data requirements can be burdensome"},{"startTime":2125.11,"endTime":2127.66,"body":"and requires a sufficient level of data"},{"startTime":2127.66,"endTime":2130.27,"body":"to be collected from a clinical trial."},{"startTime":2130.27,"endTime":2132.43,"body":"Also, if you're looking"},{"startTime":2130.27,"endTime":2132.43,"body":"at specific tunnel states,"},{"startTime":2132.43,"endTime":2134.95,"body":"you've got to apply a cost, utility"},{"startTime":2134.95,"endTime":2136.84,"body":"and resource use to those states."},{"startTime":2136.84,"endTime":2139.12,"body":"So it'll be important"},{"startTime":2136.84,"endTime":2139.12,"body":"to identify for example"},{"startTime":2139.12,"endTime":2141.97,"body":"through a literature search"},{"startTime":2139.12,"endTime":2141.97,"body":"data which is most appropriate"},{"startTime":2141.97,"endTime":2143.57,"body":"for that specific disease state."},{"startTime":2144.602,"endTime":2147.206,"body":"The interpretability and"},{"startTime":2144.602,"endTime":2147.206,"body":"transparency of the states."},{"startTime":2147.206,"endTime":2150.28,"body":"As we were discussing"},{"startTime":2147.206,"endTime":2150.28,"body":"with the decision trees,"},{"startTime":2150.28,"endTime":2152.26,"body":"when you have quite a lot of say branches"},{"startTime":2152.26,"endTime":2155.022,"body":"in that instance it can be"},{"startTime":2152.26,"endTime":2155.022,"body":"quite difficult to communicate"},{"startTime":2155.022,"endTime":2159.88,"body":"the differences and it can"},{"startTime":2155.022,"endTime":2159.88,"body":"also look quite messy at times."},{"startTime":2159.88,"endTime":2161.11,"body":"That's also reflective here."},{"startTime":2161.11,"endTime":2163.99,"body":"If we have multiple health"},{"startTime":2161.11,"endTime":2163.99,"body":"states, it can be quite complex"},{"startTime":2163.99,"endTime":2165.49,"body":"to discuss for different transitions"},{"startTime":2165.49,"endTime":2167.19,"body":"that can occur between the states."},{"startTime":2168.64,"endTime":2171.7,"body":"The computational demand"},{"startTime":2168.64,"endTime":2171.7,"body":"should also be considered here."},{"startTime":2171.7,"endTime":2175.125,"body":"So in Excel, if you're building"},{"startTime":2171.7,"endTime":2175.125,"body":"a Markov model for example,"},{"startTime":2175.125,"endTime":2177.3,"body":"you'd usually have three,"},{"startTime":2177.3,"endTime":2180.042,"body":"say so for example, in this instance,"},{"startTime":2180.042,"endTime":2180.97,"body":"you'd have three health states"},{"startTime":2180.97,"endTime":2182.89,"body":"which can be corresponding"},{"startTime":2180.97,"endTime":2182.89,"body":"into the model engine"},{"startTime":2182.89,"endTime":2184.0,"body":"which should have three states,"},{"startTime":2184.0,"endTime":2186.763,"body":"three lots of transitions to and from."},{"startTime":2187.64,"endTime":2190.7,"body":"For example, in this case"},{"startTime":2187.64,"endTime":2190.7,"body":"if we have nine states,"},{"startTime":2190.7,"endTime":2194.566,"body":"that triples the sort of"},{"startTime":2190.7,"endTime":2194.566,"body":"burden computationally"},{"startTime":2194.566,"endTime":2198.819,"body":"in the model, which can add"},{"startTime":2194.566,"endTime":2198.819,"body":"to the size the Excel model"},{"startTime":2198.819,"endTime":2201.22,"body":"and add to the sort of difficulty."},{"startTime":2201.22,"endTime":2203.44,"body":"Or not difficulty, I'd"},{"startTime":2201.22,"endTime":2203.44,"body":"say add to the speed."},{"startTime":2203.44,"endTime":2205.984,"body":"So for example, if you're"},{"startTime":2203.44,"endTime":2205.984,"body":"running simulations later on,"},{"startTime":2205.984,"endTime":2207.857,"body":"it can add to the time"},{"startTime":2207.857,"endTime":2209.95,"body":"because you want to go"},{"startTime":2207.857,"endTime":2209.95,"body":"through triple the amount"},{"startTime":2209.95,"endTime":2212.353,"body":"of data and movement."},{"startTime":2214.24,"endTime":2217.09,"body":"So for the next model"},{"startTime":2214.24,"endTime":2217.09,"body":"concept, Hannah's going"},{"startTime":2217.09,"endTime":2221.745,"body":"to be discussing the"},{"startTime":2217.09,"endTime":2221.745,"body":"discrete event simulation,"},{"startTime":2221.745,"endTime":2224.438,"body":"which starts to move on from that sort of"},{"startTime":2224.438,"endTime":2226.54,"body":"cohort-level approach"},{"startTime":2226.54,"endTime":2228.55,"body":"and starts specifically"},{"startTime":2226.54,"endTime":2228.55,"body":"looking at patient level."},{"startTime":2228.55,"endTime":2233.433,"body":"So specific events and"},{"startTime":2228.55,"endTime":2233.433,"body":"simulations that occur."},{"startTime":2236.663,"endTime":2238.939,"body":"- Yeah, okay great, thank you."},{"startTime":2238.939,"endTime":2241.36,"body":"So discrete event simulations are one"},{"startTime":2241.36,"endTime":2242.98,"body":"of my personal favourites"},{"startTime":2242.98,"endTime":2245.92,"body":"and they represent one way"},{"startTime":2242.98,"endTime":2245.92,"body":"of overcoming the challenges"},{"startTime":2245.92,"endTime":2248.11,"body":"of Markov or semi-Markov models"},{"startTime":2248.11,"endTime":2250.832,"body":"when the time dependency"},{"startTime":2248.11,"endTime":2250.832,"body":"is really important."},{"startTime":2250.832,"endTime":2254.56,"body":"A discrete event simulation is inherently"},{"startTime":2254.56,"endTime":2257.62,"body":"a patient-level model because"},{"startTime":2254.56,"endTime":2257.62,"body":"of the way that it's set up"},{"startTime":2257.62,"endTime":2260.263,"body":"and its structure, which"},{"startTime":2257.62,"endTime":2260.263,"body":"we will get into now."},{"startTime":2262.21,"endTime":2265.39,"body":"So a discrete event simulation is a type"},{"startTime":2265.39,"endTime":2267.921,"body":"of patient-level simulation model."},{"startTime":2267.921,"endTime":2271.155,"body":"The time to events are"},{"startTime":2267.921,"endTime":2271.155,"body":"sampled in discrete time,"},{"startTime":2271.155,"endTime":2273.79,"body":"which can be more realistic"},{"startTime":2271.155,"endTime":2273.79,"body":"than moving patients"},{"startTime":2273.79,"endTime":2275.86,"body":"through the model in set cycles"},{"startTime":2275.86,"endTime":2278.19,"body":"as you would in a Markov model."},{"startTime":2278.19,"endTime":2280.99,"body":"So if I talk through the"},{"startTime":2278.19,"endTime":2280.99,"body":"example model schematic"},{"startTime":2280.99,"endTime":2284.14,"body":"on the right hand side, this"},{"startTime":2280.99,"endTime":2284.14,"body":"will help better explain"},{"startTime":2284.14,"endTime":2287.203,"body":"how the discrete event"},{"startTime":2284.14,"endTime":2287.203,"body":"simulation works computationally."},{"startTime":2288.91,"endTime":2292.09,"body":"So the model starts by"},{"startTime":2288.91,"endTime":2292.09,"body":"sampling one patient"},{"startTime":2292.09,"endTime":2294.7,"body":"from individual patient-level data."},{"startTime":2294.7,"endTime":2297.46,"body":"So you take that one patient's data points"},{"startTime":2297.46,"endTime":2299.5,"body":"and their characteristics are stored."},{"startTime":2299.5,"endTime":2302.41,"body":"So this can be things like their age"},{"startTime":2302.41,"endTime":2306.328,"body":"other characteristics related"},{"startTime":2302.41,"endTime":2306.328,"body":"to the disease, their gender."},{"startTime":2306.328,"endTime":2309.373,"body":"And then based on these characteristics"},{"startTime":2309.373,"endTime":2312.01,"body":"for that one patient, the time"},{"startTime":2312.01,"endTime":2314.47,"body":"to different events are estimated."},{"startTime":2314.47,"endTime":2317.29,"body":"This is usually based"},{"startTime":2314.47,"endTime":2317.29,"body":"on regression equations."},{"startTime":2317.29,"endTime":2320.56,"body":"So these are predictive"},{"startTime":2317.29,"endTime":2320.56,"body":"equations that take inputs"},{"startTime":2320.56,"endTime":2322.3,"body":"such as the patient characteristics"},{"startTime":2322.3,"endTime":2324.048,"body":"and predict an output"},{"startTime":2324.048,"endTime":2329.048,"body":"based on those characteristics."},{"startTime":2329.59,"endTime":2331.806,"body":"And in this example, the"},{"startTime":2329.59,"endTime":2331.806,"body":"model estimated the time"},{"startTime":2331.806,"endTime":2336.806,"body":"to discontinuing treatment,"},{"startTime":2331.806,"endTime":2336.806,"body":"the time to a change in CD4,"},{"startTime":2336.97,"endTime":2340.903,"body":"which is a measure used in"},{"startTime":2336.97,"endTime":2340.903,"body":"HIV, and the time to death."},{"startTime":2341.92,"endTime":2344.754,"body":"And then whichever of these"},{"startTime":2341.92,"endTime":2344.754,"body":"three times is the shortest,"},{"startTime":2344.754,"endTime":2349.003,"body":"this event is determined to be"},{"startTime":2344.754,"endTime":2349.003,"body":"the next event that happens."},{"startTime":2349.854,"endTime":2352.33,"body":"Then the relevant costs"},{"startTime":2352.33,"endTime":2355.78,"body":"and QALYs are calculated based on this"},{"startTime":2355.78,"endTime":2358.0,"body":"and the amount of time that has passed."},{"startTime":2358.0,"endTime":2362.05,"body":"So whereas before in for"},{"startTime":2358.0,"endTime":2362.05,"body":"example a Markov model,"},{"startTime":2362.05,"endTime":2366.755,"body":"the costs and QALYs would"},{"startTime":2362.05,"endTime":2366.755,"body":"always be based on the same,"},{"startTime":2366.755,"endTime":2368.8,"body":"for example, monthly cycle."},{"startTime":2368.8,"endTime":2371.47,"body":"Here you have to take into"},{"startTime":2368.8,"endTime":2371.47,"body":"account the time that has passed"},{"startTime":2371.47,"endTime":2374.783,"body":"because the time to event can be changing"},{"startTime":2374.783,"endTime":2376.873,"body":"every time it's estimated."},{"startTime":2377.726,"endTime":2381.58,"body":"So then if the patient"},{"startTime":2377.726,"endTime":2381.58,"body":"dies, they exit the model."},{"startTime":2381.58,"endTime":2385.06,"body":"If they discontinue, they"},{"startTime":2381.58,"endTime":2385.06,"body":"switch to the next treatment."},{"startTime":2385.06,"endTime":2388.27,"body":"And this is useful to know"},{"startTime":2388.27,"endTime":2391.54,"body":"if you want to model treatment sequences,"},{"startTime":2391.54,"endTime":2393.79,"body":"then a discrete event"},{"startTime":2391.54,"endTime":2393.79,"body":"simulation is a great way"},{"startTime":2393.79,"endTime":2395.953,"body":"to be able to incorporate"},{"startTime":2393.79,"endTime":2395.953,"body":"that flexibility."},{"startTime":2397.33,"endTime":2400.21,"body":"And then if a CD4 change occurs,"},{"startTime":2400.21,"endTime":2404.029,"body":"then the times are all re-estimated"},{"startTime":2404.029,"endTime":2406.066,"body":"and the process continues"},{"startTime":2406.066,"endTime":2408.703,"body":"until the patient has exited the model."},{"startTime":2410.304,"endTime":2414.408,"body":"Then, every time a new event is estimated,"},{"startTime":2414.408,"endTime":2416.65,"body":"it will depend on the time elapsed,"},{"startTime":2416.65,"endTime":2418.03,"body":"the patient's characteristics,"},{"startTime":2418.03,"endTime":2419.83,"body":"which can be updated over time."},{"startTime":2419.83,"endTime":2422.62,"body":"So their age will increase accordingly."},{"startTime":2422.62,"endTime":2425.83,"body":"If an event has happened that"},{"startTime":2422.62,"endTime":2425.83,"body":"can potentially be recorded"},{"startTime":2425.83,"endTime":2427.63,"body":"and the information can be used"},{"startTime":2427.63,"endTime":2430.39,"body":"to predict what happens next."},{"startTime":2430.39,"endTime":2433.03,"body":"And this is then, this whole"},{"startTime":2430.39,"endTime":2433.03,"body":"process is then repeated,"},{"startTime":2433.03,"endTime":2437.26,"body":"for every patient that"},{"startTime":2433.03,"endTime":2437.26,"body":"goes through the model."},{"startTime":2437.26,"endTime":2438.969,"body":"So using the individual patient data"},{"startTime":2438.969,"endTime":2442.813,"body":"and then is all aggregated at the end."},{"startTime":2444.22,"endTime":2448.39,"body":"So just to summarise that again,"},{"startTime":2448.39,"endTime":2451.69,"body":"so the event likelihoods can be determined"},{"startTime":2451.69,"endTime":2454.72,"body":"by individual patient characteristics."},{"startTime":2454.72,"endTime":2456.61,"body":"These are recorded at baseline"},{"startTime":2456.61,"endTime":2458.86,"body":"and they can be updated over time."},{"startTime":2458.86,"endTime":2460.57,"body":"The costs and QALYs are estimated"},{"startTime":2460.57,"endTime":2463.3,"body":"and stored when an event happens."},{"startTime":2463.3,"endTime":2465.7,"body":"And this takes into"},{"startTime":2463.3,"endTime":2465.7,"body":"consideration the amount of time"},{"startTime":2465.7,"endTime":2468.01,"body":"that has passed since the last event."},{"startTime":2468.01,"endTime":2470.381,"body":"The results are then aggregated over time"},{"startTime":2470.381,"endTime":2472.42,"body":"and for all the patients,"},{"startTime":2472.42,"endTime":2475.663,"body":"to produce the summary"},{"startTime":2472.42,"endTime":2475.663,"body":"experience for the whole cohort."},{"startTime":2476.783,"endTime":2479.53,"body":"And discrete event simulation"},{"startTime":2476.783,"endTime":2479.53,"body":"is therefore likely"},{"startTime":2479.53,"endTime":2482.56,"body":"to be really useful for"},{"startTime":2479.53,"endTime":2482.56,"body":"modelling complex conditions"},{"startTime":2482.56,"endTime":2484.84,"body":"where there are many"},{"startTime":2482.56,"endTime":2484.84,"body":"different types of events"},{"startTime":2484.84,"endTime":2486.037,"body":"and health states."},{"startTime":2486.037,"endTime":2488.783,"body":"And one example of this"},{"startTime":2486.037,"endTime":2488.783,"body":"could be the complications"},{"startTime":2488.783,"endTime":2491.928,"body":"of diabetes when there"},{"startTime":2488.783,"endTime":2491.928,"body":"can be different events"},{"startTime":2491.928,"endTime":2494.92,"body":"that the patient is at risk of,"},{"startTime":2494.92,"endTime":2497.89,"body":"for example, cardiovascular"},{"startTime":2494.92,"endTime":2497.89,"body":"events such as strokes"},{"startTime":2497.89,"endTime":2501.309,"body":"and heart attacks and situations, also"},{"startTime":2501.309,"endTime":2505.123,"body":"where the patient's history"},{"startTime":2501.309,"endTime":2505.123,"body":"impacts the future events."},{"startTime":2506.48,"endTime":2509.5,"body":"So the strengths are that it can account"},{"startTime":2509.5,"endTime":2511.69,"body":"for the patient heterogeneity"},{"startTime":2511.69,"endTime":2513.49,"body":"with it being a patient-level model."},{"startTime":2514.36,"endTime":2518.779,"body":"The flexibility can be applied"},{"startTime":2514.36,"endTime":2518.779,"body":"from modelling multiple events"},{"startTime":2518.779,"endTime":2522.463,"body":"and also for taking into"},{"startTime":2518.779,"endTime":2522.463,"body":"account the time dependency."},{"startTime":2523.51,"endTime":2527.05,"body":"However, the main limitation"},{"startTime":2523.51,"endTime":2527.05,"body":"is that it does rely heavily"},{"startTime":2527.05,"endTime":2529.72,"body":"on data and is more complex."},{"startTime":2529.72,"endTime":2532.0,"body":"So whereas the other"},{"startTime":2529.72,"endTime":2532.0,"body":"models we've discussed"},{"startTime":2532.0,"endTime":2535.36,"body":"so far could be typically"},{"startTime":2532.0,"endTime":2535.36,"body":"developed in Excel,"},{"startTime":2535.36,"endTime":2537.49,"body":"a discrete event simulation would need"},{"startTime":2537.49,"endTime":2539.564,"body":"to use a programming language."},{"startTime":2539.564,"endTime":2542.98,"body":"One way to do this is to use Visual Basic"},{"startTime":2542.98,"endTime":2546.67,"body":"or VBA, which is Excel's"},{"startTime":2542.98,"endTime":2546.67,"body":"inbuilt programming language."},{"startTime":2546.67,"endTime":2549.64,"body":"And this allows for Excel to still be used"},{"startTime":2549.64,"endTime":2552.577,"body":"as an interface for the inputs"},{"startTime":2552.577,"endTime":2556.783,"body":"and the VBA programming is"},{"startTime":2552.577,"endTime":2556.783,"body":"being used in the background."},{"startTime":2557.958,"endTime":2561.28,"body":"The discrete event"},{"startTime":2557.958,"endTime":2561.28,"body":"simulation does rely heavily"},{"startTime":2561.28,"endTime":2563.35,"body":"on having access to patient-level data"},{"startTime":2563.35,"endTime":2565.524,"body":"and that being available"},{"startTime":2563.35,"endTime":2565.524,"body":"to use in the model."},{"startTime":2565.524,"endTime":2567.19,"body":"And because of that,"},{"startTime":2567.19,"endTime":2571.93,"body":"it can sometimes limit this approach."},{"startTime":2571.93,"endTime":2574.463,"body":"So if the patient-level"},{"startTime":2571.93,"endTime":2574.463,"body":"data isn't available,"},{"startTime":2574.463,"endTime":2577.12,"body":"then another alternative might be needed."},{"startTime":2577.12,"endTime":2579.37,"body":"And one of these that we wanted"},{"startTime":2579.37,"endTime":2582.52,"body":"to discuss today is the"},{"startTime":2579.37,"endTime":2582.52,"body":"Cox regression approach,"},{"startTime":2582.52,"endTime":2585.12,"body":"which I will hand over to"},{"startTime":2582.52,"endTime":2585.12,"body":"Danny to walk you through."},{"startTime":2586.33,"endTime":2587.437,"body":"- Thanks Hannah."},{"startTime":2587.437,"endTime":2588.958,"body":"So as discussed,"},{"startTime":2588.958,"endTime":2593.47,"body":"the Cox regression approach"},{"startTime":2588.958,"endTime":2593.47,"body":"could be a alternative method"},{"startTime":2593.47,"endTime":2595.872,"body":"used to discrete event simulation."},{"startTime":2595.872,"endTime":2599.802,"body":"So on to the next slide, thank you."},{"startTime":2599.802,"endTime":2602.55,"body":"So a Cox regression model, I'd say,"},{"startTime":2602.55,"endTime":2606.324,"body":"is not incredibly commonplace"},{"startTime":2602.55,"endTime":2606.324,"body":"in HTA submissions."},{"startTime":2606.324,"endTime":2610.18,"body":"There isn't usually a"},{"startTime":2606.324,"endTime":2610.18,"body":"fantastic precedent out there,"},{"startTime":2610.18,"endTime":2611.74,"body":"but it's a very intuitive"},{"startTime":2611.74,"endTime":2614.35,"body":"and interesting approach to consider."},{"startTime":2614.35,"endTime":2616.764,"body":"From my personal experience,"},{"startTime":2616.764,"endTime":2620.26,"body":"for a previous submission"},{"startTime":2616.764,"endTime":2620.26,"body":"I worked on a model"},{"startTime":2620.26,"endTime":2622.3,"body":"which incorporated a"},{"startTime":2620.26,"endTime":2622.3,"body":"Cox regression approach"},{"startTime":2622.3,"endTime":2623.92,"body":"and it was very useful"},{"startTime":2623.92,"endTime":2627.48,"body":"and logical let's say"},{"startTime":2623.92,"endTime":2627.48,"body":"in incorporating events"},{"startTime":2627.48,"endTime":2630.719,"body":"in the specific indication."},{"startTime":2630.719,"endTime":2633.52,"body":"So for example, if we're talking"},{"startTime":2633.52,"endTime":2637.12,"body":"about say a cardiovascular indication"},{"startTime":2637.12,"endTime":2639.85,"body":"where you've got significant"},{"startTime":2637.12,"endTime":2639.85,"body":"amount of cardiovascular events"},{"startTime":2639.85,"endTime":2641.5,"body":"which should be incorporated,"},{"startTime":2641.5,"endTime":2643.9,"body":"the Cox regression"},{"startTime":2641.5,"endTime":2643.9,"body":"approach would allow you"},{"startTime":2643.9,"endTime":2646.69,"body":"to incorporate the different events"},{"startTime":2646.69,"endTime":2651.07,"body":"and use them to estimate the"},{"startTime":2646.69,"endTime":2651.07,"body":"likelihood for specific events."},{"startTime":2651.07,"endTime":2655.033,"body":"So for example, death, the"},{"startTime":2651.07,"endTime":2655.033,"body":"likelihood of that occurring."},{"startTime":2656.89,"endTime":2659.86,"body":"So with those cardiovascular outcomes,"},{"startTime":2659.86,"endTime":2662.56,"body":"an example, if we take heart failure,"},{"startTime":2662.56,"endTime":2665.59,"body":"what the Cox regression approach"},{"startTime":2662.56,"endTime":2665.59,"body":"does is it takes a series"},{"startTime":2665.59,"endTime":2669.677,"body":"of covariates, so in a regression"},{"startTime":2665.59,"endTime":2669.677,"body":"sort of form of equation"},{"startTime":2669.677,"endTime":2672.633,"body":"and uses it to estimate the likelihood"},{"startTime":2672.633,"endTime":2675.793,"body":"of a heart failure event occurring."},{"startTime":2676.9,"endTime":2678.64,"body":"In the example that I've used,"},{"startTime":2678.64,"endTime":2680.23,"body":"we use this for several events."},{"startTime":2680.23,"endTime":2682.395,"body":"So we worked out, for"},{"startTime":2680.23,"endTime":2682.395,"body":"example, heart failure"},{"startTime":2682.395,"endTime":2685.87,"body":"and cardiovascular-related mortality."},{"startTime":2685.87,"endTime":2687.64,"body":"We were able to work out the likelihood"},{"startTime":2687.64,"endTime":2689.17,"body":"of each event occurring"},{"startTime":2689.17,"endTime":2690.49,"body":"and then what we're able to do"},{"startTime":2690.49,"endTime":2693.018,"body":"is estimate the cumulative hazard."},{"startTime":2693.018,"endTime":2695.579,"body":"The hazard being the"},{"startTime":2693.018,"endTime":2695.579,"body":"likelihood of an event"},{"startTime":2695.579,"endTime":2699.49,"body":"in an intervention group"},{"startTime":2695.579,"endTime":2699.49,"body":"relative to the comparator."},{"startTime":2699.49,"endTime":2702.55,"body":"And using that we were able"},{"startTime":2699.49,"endTime":2702.55,"body":"to estimate transitions"},{"startTime":2702.55,"endTime":2703.996,"body":"for patients over time."},{"startTime":2703.996,"endTime":2707.14,"body":"We're able to estimate the"},{"startTime":2703.996,"endTime":2707.14,"body":"proportion of patients dying"},{"startTime":2707.14,"endTime":2709.27,"body":"'cause we're able to work out okay,"},{"startTime":2709.27,"endTime":2712.03,"body":"what are the impacts of"},{"startTime":2709.27,"endTime":2712.03,"body":"the individual covariates"},{"startTime":2712.03,"endTime":2715.18,"body":"or so for example, age or demographic"},{"startTime":2715.18,"endTime":2717.49,"body":"or smoker status for example,"},{"startTime":2717.49,"endTime":2720.01,"body":"what's the impact of"},{"startTime":2717.49,"endTime":2720.01,"body":"those different factors"},{"startTime":2720.01,"endTime":2722.86,"body":"on the likelihood of an"},{"startTime":2720.01,"endTime":2722.86,"body":"event of death taking place?"},{"startTime":2722.86,"endTime":2724.94,"body":"We could then incorporate"},{"startTime":2722.86,"endTime":2724.94,"body":"that into a model"},{"startTime":2724.94,"endTime":2729.94,"body":"sort of matrix format to"},{"startTime":2724.94,"endTime":2729.94,"body":"estimate okay, what the number"},{"startTime":2730.396,"endTime":2733.717,"body":"of patients done at each stage were."},{"startTime":2733.717,"endTime":2736.57,"body":"We were also able to incorporate"},{"startTime":2733.717,"endTime":2736.57,"body":"the different events,"},{"startTime":2736.57,"endTime":2739.54,"body":"so the likelihood of a"},{"startTime":2736.57,"endTime":2739.54,"body":"patient having heart failure"},{"startTime":2739.54,"endTime":2743.38,"body":"for example, and we're able"},{"startTime":2739.54,"endTime":2743.38,"body":"to identify costs for that."},{"startTime":2743.38,"endTime":2745.09,"body":"And we're able to model, okay,"},{"startTime":2745.09,"endTime":2747.777,"body":"so how many patients are"},{"startTime":2745.09,"endTime":2747.777,"body":"experiencing a number of events,"},{"startTime":2747.777,"endTime":2751.6,"body":"can we assign a cost to"},{"startTime":2747.777,"endTime":2751.6,"body":"that, a utility to that?"},{"startTime":2751.6,"endTime":2755.56,"body":"And it allowed a reflective"},{"startTime":2751.6,"endTime":2755.56,"body":"modelling process"},{"startTime":2755.56,"endTime":2758.38,"body":"to be undertaken for"},{"startTime":2755.56,"endTime":2758.38,"body":"that specific indication,"},{"startTime":2758.38,"endTime":2760.13,"body":"looking at cardiovascular outcomes."},{"startTime":2761.38,"endTime":2763.75,"body":"There's various strengths"},{"startTime":2761.38,"endTime":2763.75,"body":"to the modelling approach."},{"startTime":2763.75,"endTime":2765.79,"body":"So it allows you to"},{"startTime":2763.75,"endTime":2765.79,"body":"look at the association"},{"startTime":2765.79,"endTime":2767.65,"body":"between the survival time of patients"},{"startTime":2767.65,"endTime":2770.11,"body":"and one more predictive"},{"startTime":2767.65,"endTime":2770.11,"body":"variables to assess,"},{"startTime":2770.11,"endTime":2771.58,"body":"okay, what's the likelihood,"},{"startTime":2771.58,"endTime":2773.14,"body":"what happens if we increase the likelihood"},{"startTime":2773.14,"endTime":2774.49,"body":"of one variable having,"},{"startTime":2774.49,"endTime":2776.477,"body":"what impact will that have on survival?"},{"startTime":2776.477,"endTime":2778.303,"body":"It allows the incorporation of patients"},{"startTime":2778.303,"endTime":2780.97,"body":"that are yet to experience an event."},{"startTime":2780.97,"endTime":2784.69,"body":"We're able to simulate"},{"startTime":2780.97,"endTime":2784.69,"body":"basically what the impact"},{"startTime":2784.69,"endTime":2787.33,"body":"of each varying each covariate would be"},{"startTime":2787.33,"endTime":2788.95,"body":"on the likelihood of survival"},{"startTime":2788.95,"endTime":2791.44,"body":"and use that to incorporate"},{"startTime":2788.95,"endTime":2791.44,"body":"future patients,"},{"startTime":2791.44,"endTime":2793.308,"body":"the expectation of what"},{"startTime":2791.44,"endTime":2793.308,"body":"would happen to them"},{"startTime":2793.308,"endTime":2796.063,"body":"on the likelihood of"},{"startTime":2793.308,"endTime":2796.063,"body":"them receiving an event."},{"startTime":2796.93,"endTime":2798.631,"body":"There are a couple of limitations."},{"startTime":2798.631,"endTime":2802.338,"body":"So it doesn't incorporate the"},{"startTime":2798.631,"endTime":2802.338,"body":"changes in risk over time."},{"startTime":2802.338,"endTime":2805.51,"body":"So with the hazard rate, we"},{"startTime":2802.338,"endTime":2805.51,"body":"assume what hazard ratio,"},{"startTime":2805.51,"endTime":2808.48,"body":"we assume that patients"},{"startTime":2805.51,"endTime":2808.48,"body":"experience an event"},{"startTime":2808.48,"endTime":2812.02,"body":"with a fixed likelihood"},{"startTime":2808.48,"endTime":2812.02,"body":"in an intervention group"},{"startTime":2812.02,"endTime":2814.21,"body":"relative to a comparator."},{"startTime":2814.21,"endTime":2816.88,"body":"An example being a hazard ratio of 1.5,"},{"startTime":2816.88,"endTime":2821.02,"body":"would assume a 50% increase"},{"startTime":2816.88,"endTime":2821.02,"body":"in the likelihood of an event"},{"startTime":2821.02,"endTime":2823.54,"body":"in the intervention arm"},{"startTime":2821.02,"endTime":2823.54,"body":"relative to the comparator."},{"startTime":2823.54,"endTime":2825.802,"body":"This won't change over time in the model."},{"startTime":2825.802,"endTime":2828.22,"body":"The assumption of proportional hazards"},{"startTime":2828.22,"endTime":2829.36,"body":"is often not realistic."},{"startTime":2829.36,"endTime":2832.0,"body":"So proportional hazards"},{"startTime":2829.36,"endTime":2832.0,"body":"looks at the hazard rates,"},{"startTime":2832.0,"endTime":2834.85,"body":"the likelihood of an event"},{"startTime":2832.0,"endTime":2834.85,"body":"between an intervention"},{"startTime":2834.85,"endTime":2836.23,"body":"and a control group"},{"startTime":2836.23,"endTime":2839.2,"body":"and it assumes that it remains"},{"startTime":2836.23,"endTime":2839.2,"body":"proportional over time."},{"startTime":2839.2,"endTime":2841.53,"body":"Often the hazard rates at the likelihood"},{"startTime":2841.53,"endTime":2846.16,"body":"of an event occurring can"},{"startTime":2841.53,"endTime":2846.16,"body":"vary in a intervention group,"},{"startTime":2846.16,"endTime":2848.383,"body":"say relative to a control arm."},{"startTime":2848.383,"endTime":2852.64,"body":"And this assumption isn't"},{"startTime":2848.383,"endTime":2852.64,"body":"particularly realistic always"},{"startTime":2852.64,"endTime":2854.29,"body":"with the Cox regression approach."},{"startTime":2855.594,"endTime":2858.16,"body":"The final approach we're"},{"startTime":2855.594,"endTime":2858.16,"body":"going to be discussing,"},{"startTime":2858.16,"endTime":2860.11,"body":"it's one that specifically looks at,"},{"startTime":2860.11,"endTime":2863.23,"body":"I'd say specifically looks"},{"startTime":2860.11,"endTime":2863.23,"body":"at oncology modelling."},{"startTime":2863.23,"endTime":2865.3,"body":"It's very commonplace used there."},{"startTime":2865.3,"endTime":2867.64,"body":"It's the partitioned survival model"},{"startTime":2867.64,"endTime":2870.673,"body":"which Hannah's going to discuss further."},{"startTime":2873.403,"endTime":2875.115,"body":"- Yeah, thank you very much."},{"startTime":2875.115,"endTime":2879.191,"body":"So yeah, the partitioned"},{"startTime":2875.115,"endTime":2879.191,"body":"survival model is used"},{"startTime":2879.191,"endTime":2882.48,"body":"very commonly within oncology modelling."},{"startTime":2882.48,"endTime":2884.86,"body":"So for that reason you"},{"startTime":2882.48,"endTime":2884.86,"body":"may be familiar with it,"},{"startTime":2884.86,"endTime":2887.29,"body":"some of you listening today,"},{"startTime":2887.29,"endTime":2890.08,"body":"and it is inherently a cohort approach"},{"startTime":2890.08,"endTime":2891.696,"body":"because as the name suggests,"},{"startTime":2891.696,"endTime":2893.89,"body":"the whole population are partitioned"},{"startTime":2893.89,"endTime":2896.478,"body":"into different health states."},{"startTime":2896.478,"endTime":2899.77,"body":"So it is worth mentioning that"},{"startTime":2896.478,"endTime":2899.77,"body":"it can also be referred to"},{"startTime":2899.77,"endTime":2903.077,"body":"as an area under the"},{"startTime":2899.77,"endTime":2903.077,"body":"curve approach as well."},{"startTime":2903.077,"endTime":2907.75,"body":"And what happens in a"},{"startTime":2903.077,"endTime":2907.75,"body":"partitioned survival model"},{"startTime":2907.75,"endTime":2910.0,"body":"is that a cohort are followed"},{"startTime":2907.75,"endTime":2910.0,"body":"through time as they move"},{"startTime":2910.0,"endTime":2911.89,"body":"between a set of mutually exclusive"},{"startTime":2911.89,"endTime":2913.66,"body":"and exhaustive health states,"},{"startTime":2913.66,"endTime":2917.02,"body":"which sounds familiar"},{"startTime":2913.66,"endTime":2917.02,"body":"from the Markov model."},{"startTime":2917.02,"endTime":2919.42,"body":"However, the difference"},{"startTime":2917.02,"endTime":2919.42,"body":"versus a Markov model is"},{"startTime":2919.42,"endTime":2922.12,"body":"that the partitioned"},{"startTime":2919.42,"endTime":2922.12,"body":"survival model estimates"},{"startTime":2922.12,"endTime":2927.12,"body":"the proportions in each"},{"startTime":2922.12,"endTime":2927.12,"body":"state based on a set"},{"startTime":2927.19,"endTime":2929.583,"body":"of parametric survival equations."},{"startTime":2929.583,"endTime":2932.89,"body":"So one of the reasons that"},{"startTime":2929.583,"endTime":2932.89,"body":"it is so frequently used"},{"startTime":2932.89,"endTime":2936.323,"body":"in oncology is because it"},{"startTime":2932.89,"endTime":2936.323,"body":"can utilise the endpoints"},{"startTime":2936.323,"endTime":2938.62,"body":"that are commonly collected"},{"startTime":2936.323,"endTime":2938.62,"body":"in clinical trials"},{"startTime":2938.62,"endTime":2942.55,"body":"such as overall survival and"},{"startTime":2938.62,"endTime":2942.55,"body":"partitioned survival curves."},{"startTime":2942.55,"endTime":2945.22,"body":"So if you look to the example"},{"startTime":2942.55,"endTime":2945.22,"body":"on the right hand side,"},{"startTime":2945.22,"endTime":2947.29,"body":"this is exactly what we've shown here."},{"startTime":2947.29,"endTime":2949.6,"body":"So using the overall survival"},{"startTime":2949.6,"endTime":2951.67,"body":"and progression-free survival curves,"},{"startTime":2951.67,"endTime":2954.82,"body":"the cohort are partitioned"},{"startTime":2951.67,"endTime":2954.82,"body":"into three sections"},{"startTime":2954.82,"endTime":2956.35,"body":"as you can see from the colour coding."},{"startTime":2956.35,"endTime":2959.65,"body":"So, above the overall survival"},{"startTime":2956.35,"endTime":2959.65,"body":"curve are the patients"},{"startTime":2959.65,"endTime":2963.79,"body":"who have died and underneath"},{"startTime":2959.65,"endTime":2963.79,"body":"are those who are still alive."},{"startTime":2963.79,"endTime":2965.89,"body":"And then of the patients"},{"startTime":2963.79,"endTime":2965.89,"body":"who are still alive,"},{"startTime":2965.89,"endTime":2970.346,"body":"this is further proportioned"},{"startTime":2965.89,"endTime":2970.346,"body":"into two separate sections."},{"startTime":2970.346,"endTime":2975.346,"body":"Underneath the PFS curve is"},{"startTime":2970.346,"endTime":2975.346,"body":"the progression-free patients"},{"startTime":2976.087,"endTime":2980.047,"body":"and the difference between"},{"startTime":2976.087,"endTime":2980.047,"body":"the overall survival curve"},{"startTime":2980.047,"endTime":2982.72,"body":"and the progression-free"},{"startTime":2980.047,"endTime":2982.72,"body":"survival curve gives you"},{"startTime":2982.72,"endTime":2985.303,"body":"the patients who are"},{"startTime":2982.72,"endTime":2985.303,"body":"alive and have progressed."},{"startTime":2986.71,"endTime":2991.12,"body":"And, so this is a very simple"},{"startTime":2986.71,"endTime":2991.12,"body":"way of using these two curves"},{"startTime":2991.12,"endTime":2995.38,"body":"to partition the cohort"},{"startTime":2991.12,"endTime":2995.38,"body":"into these health states."},{"startTime":2995.38,"endTime":2997.78,"body":"And what we would also"},{"startTime":2995.38,"endTime":2997.78,"body":"typically need to do"},{"startTime":2997.78,"endTime":3000.06,"body":"is to extrapolate the curve."},{"startTime":3000.06,"endTime":3004.14,"body":"So these data would be"},{"startTime":3000.06,"endTime":3004.14,"body":"available from a clinical trial"},{"startTime":3004.14,"endTime":3008.484,"body":"which may for example only"},{"startTime":3004.14,"endTime":3008.484,"body":"be for one year duration."},{"startTime":3008.484,"endTime":3011.305,"body":"And then we extrapolate these to be able"},{"startTime":3011.305,"endTime":3013.68,"body":"to estimate the long-term health effects"},{"startTime":3013.68,"endTime":3015.265,"body":"of the intervention."},{"startTime":3015.265,"endTime":3018.707,"body":"What we would do is use"},{"startTime":3015.265,"endTime":3018.707,"body":"different distributions"},{"startTime":3018.707,"endTime":3020.4,"body":"and have options available."},{"startTime":3020.4,"endTime":3023.34,"body":"So these would typically be"},{"startTime":3020.4,"endTime":3023.34,"body":"Weibull, Gompertz, gamma,"},{"startTime":3023.34,"endTime":3025.653,"body":"log-logistic and log-normal."},{"startTime":3025.653,"endTime":3029.25,"body":"Survival modelling, again,"},{"startTime":3025.653,"endTime":3029.25,"body":"is a different topic,"},{"startTime":3029.25,"endTime":3030.99,"body":"but just to give you an overview"},{"startTime":3030.99,"endTime":3033.04,"body":"of that's the approach that's taken here."},{"startTime":3033.877,"endTime":3037.05,"body":"So the strengths are that it's widely used"},{"startTime":3037.05,"endTime":3038.58,"body":"in oncology modelling"},{"startTime":3038.58,"endTime":3041.1,"body":"and allows the oncology-focused endpoints"},{"startTime":3041.1,"endTime":3044.94,"body":"to be easily incorporated so"},{"startTime":3041.1,"endTime":3044.94,"body":"that the data from the trial,"},{"startTime":3044.94,"endTime":3046.65,"body":"which typically is the overall survival"},{"startTime":3046.65,"endTime":3048.93,"body":"and the progression-free"},{"startTime":3046.65,"endTime":3048.93,"body":"survival curves can directly"},{"startTime":3048.93,"endTime":3052.004,"body":"be used, which is the the main strength."},{"startTime":3052.004,"endTime":3053.804,"body":"The limitations are,"},{"startTime":3053.804,"endTime":3056.52,"body":"and it's more of just"},{"startTime":3053.804,"endTime":3056.52,"body":"something to be aware of"},{"startTime":3056.52,"endTime":3059.76,"body":"than a limitation, is that it"},{"startTime":3056.52,"endTime":3059.76,"body":"doesn't include the ability"},{"startTime":3059.76,"endTime":3061.92,"body":"to transition to an improved health state"},{"startTime":3061.92,"endTime":3062.91,"body":"from the current state."},{"startTime":3062.91,"endTime":3065.64,"body":"So this means that in this example,"},{"startTime":3065.64,"endTime":3069.09,"body":"patients can't move backwards"},{"startTime":3065.64,"endTime":3069.09,"body":"from the progressed disease"},{"startTime":3069.09,"endTime":3072.003,"body":"into back to progression-free."},{"startTime":3072.003,"endTime":3074.999,"body":"So it's important to make"},{"startTime":3072.003,"endTime":3074.999,"body":"sure that assumption holds"},{"startTime":3074.999,"endTime":3077.55,"body":"and of course this approach can be used"},{"startTime":3077.55,"endTime":3079.65,"body":"in other disease areas as well"},{"startTime":3079.65,"endTime":3083.31,"body":"where the cohort can be partitioned"},{"startTime":3083.31,"endTime":3085.233,"body":"into different health states."},{"startTime":3086.43,"endTime":3089.73,"body":"And so now that we've described"},{"startTime":3086.43,"endTime":3089.73,"body":"these six modelling types,"},{"startTime":3089.73,"endTime":3091.59,"body":"we want to mention that it's also possible"},{"startTime":3091.59,"endTime":3093.324,"body":"to use a combination of approaches."},{"startTime":3093.324,"endTime":3096.036,"body":"And I'll hand it back to"},{"startTime":3093.324,"endTime":3096.036,"body":"Danny again to show you,"},{"startTime":3096.036,"endTime":3098.336,"body":"and talk you through how"},{"startTime":3096.036,"endTime":3098.336,"body":"that can be possible."},{"startTime":3099.42,"endTime":3100.37,"body":"- Thank you Hannah."},{"startTime":3101.328,"endTime":3103.466,"body":"So we've discussed the six sort"},{"startTime":3103.466,"endTime":3106.316,"body":"of key modelling"},{"startTime":3103.466,"endTime":3106.316,"body":"approaches and structures."},{"startTime":3106.316,"endTime":3111.316,"body":"However, hybrid modelling tries"},{"startTime":3106.316,"endTime":3111.316,"body":"to approach decision problems"},{"startTime":3111.78,"endTime":3114.69,"body":"where there may not be a"},{"startTime":3111.78,"endTime":3114.69,"body":"specific one of those six"},{"startTime":3114.69,"endTime":3116.25,"body":"which we want to look at."},{"startTime":3116.25,"endTime":3118.26,"body":"In some cases there's an argument"},{"startTime":3118.26,"endTime":3121.147,"body":"that using two structures at the same time"},{"startTime":3121.147,"endTime":3123.36,"body":"or in within a model sort"},{"startTime":3123.36,"endTime":3125.613,"body":"of overall modelling"},{"startTime":3123.36,"endTime":3125.613,"body":"process is applicable."},{"startTime":3126.69,"endTime":3129.75,"body":"So in economic modelling it is crucial"},{"startTime":3129.75,"endTime":3132.21,"body":"that the model provides"},{"startTime":3129.75,"endTime":3132.21,"body":"an accurate reflection"},{"startTime":3132.21,"endTime":3133.56,"body":"of disease characteristics"},{"startTime":3133.56,"endTime":3136.025,"body":"and the disease pathway over time."},{"startTime":3136.025,"endTime":3139.752,"body":"Examples where we may use hybrid modelling"},{"startTime":3139.752,"endTime":3143.107,"body":"are when we have say,"},{"startTime":3139.752,"endTime":3143.107,"body":"response-based models."},{"startTime":3143.107,"endTime":3147.629,"body":"So models where a patient could"},{"startTime":3143.107,"endTime":3147.629,"body":"experience a specific event"},{"startTime":3147.629,"endTime":3149.19,"body":"and then that is expected"},{"startTime":3149.19,"endTime":3152.31,"body":"to have an implication"},{"startTime":3149.19,"endTime":3152.31,"body":"on the health outcomes"},{"startTime":3152.31,"endTime":3154.41,"body":"for the remainder of the time horizon"},{"startTime":3154.41,"endTime":3157.5,"body":"or potentially the"},{"startTime":3154.41,"endTime":3157.5,"body":"remainder of their lifetime."},{"startTime":3157.5,"endTime":3159.947,"body":"So in cases such as"},{"startTime":3157.5,"endTime":3159.947,"body":"response-based modelling,"},{"startTime":3159.947,"endTime":3163.56,"body":"one of the modelling approaches"},{"startTime":3159.947,"endTime":3163.56,"body":"may not be sufficient."},{"startTime":3163.56,"endTime":3167.668,"body":"So what we may use instead is"},{"startTime":3163.56,"endTime":3167.668,"body":"say a decision tree process."},{"startTime":3167.668,"endTime":3170.01,"body":"So we'd be able to look"},{"startTime":3167.668,"endTime":3170.01,"body":"at the decision tree"},{"startTime":3170.01,"endTime":3172.23,"body":"and say, okay, in each of the branches,"},{"startTime":3172.23,"endTime":3173.85,"body":"what's the likelihood of a patient"},{"startTime":3173.85,"endTime":3176.52,"body":"experiencing a specific response status?"},{"startTime":3176.52,"endTime":3179.464,"body":"So for example, a complete"},{"startTime":3176.52,"endTime":3179.464,"body":"response or a partial response,"},{"startTime":3179.464,"endTime":3182.4,"body":"what's the probability of each"},{"startTime":3182.4,"endTime":3186.0,"body":"of those response status happening?"},{"startTime":3186.0,"endTime":3188.825,"body":"What we could then say is, okay,"},{"startTime":3188.825,"endTime":3191.25,"body":"that's defined over say a"},{"startTime":3188.825,"endTime":3191.25,"body":"month or a yearly time period."},{"startTime":3191.25,"endTime":3192.99,"body":"Can we then assume"},{"startTime":3192.99,"endTime":3195.84,"body":"that the longer term"},{"startTime":3192.99,"endTime":3195.84,"body":"outcomes can be represented"},{"startTime":3195.84,"endTime":3197.34,"body":"in say a Markov model?"},{"startTime":3197.34,"endTime":3198.48,"body":"So that could be in cases"},{"startTime":3198.48,"endTime":3200.63,"body":"where a decision tree is not appropriate"},{"startTime":3200.63,"endTime":3203.07,"body":"to model over the entire time horizon."},{"startTime":3203.07,"endTime":3206.633,"body":"Could we then assume that"},{"startTime":3203.07,"endTime":3206.633,"body":"patients go into a specific state"},{"startTime":3206.633,"endTime":3210.09,"body":"in the Markov model and"},{"startTime":3206.633,"endTime":3210.09,"body":"then have the model continue"},{"startTime":3210.09,"endTime":3212.883,"body":"with transitions in a"},{"startTime":3210.09,"endTime":3212.883,"body":"simple Markov process?"},{"startTime":3213.935,"endTime":3217.2,"body":"Hybrid modelling approaches,"},{"startTime":3213.935,"endTime":3217.2,"body":"they're not negatively viewed"},{"startTime":3217.2,"endTime":3220.954,"body":"by NICE, although structure"},{"startTime":3217.2,"endTime":3220.954,"body":"will require justification."},{"startTime":3220.954,"endTime":3224.827,"body":"As long as it's representing"},{"startTime":3220.954,"endTime":3224.827,"body":"the disease pathway"},{"startTime":3224.827,"endTime":3227.275,"body":"in an accurate and optimal way,"},{"startTime":3227.275,"endTime":3229.15,"body":"there will always be an argument"},{"startTime":3229.15,"endTime":3231.809,"body":"and it's just about being able"},{"startTime":3229.15,"endTime":3231.809,"body":"to communicate the argument"},{"startTime":3231.809,"endTime":3233.43,"body":"that the structure is adequate"},{"startTime":3233.43,"endTime":3236.345,"body":"and a good reflection"},{"startTime":3233.43,"endTime":3236.345,"body":"of a real world setting,"},{"startTime":3236.345,"endTime":3239.068,"body":"in which case it'll be most appropriate."},{"startTime":3239.068,"endTime":3241.113,"body":"So the next slide please, Hannah."},{"startTime":3242.07,"endTime":3244.847,"body":"So just building upon the"},{"startTime":3242.07,"endTime":3244.847,"body":"example that I described here,"},{"startTime":3244.847,"endTime":3247.593,"body":"if we had say a response-based model,"},{"startTime":3247.593,"endTime":3250.68,"body":"patients, we'd be able"},{"startTime":3247.593,"endTime":3250.68,"body":"to define that patients"},{"startTime":3250.68,"endTime":3252.93,"body":"for the first half a year or year"},{"startTime":3252.93,"endTime":3255.871,"body":"of a time horizon are in a"},{"startTime":3252.93,"endTime":3255.871,"body":"decision tree health state."},{"startTime":3255.871,"endTime":3258.63,"body":"Patients are allocated"},{"startTime":3255.871,"endTime":3258.63,"body":"to any of the branches"},{"startTime":3258.63,"endTime":3261.0,"body":"based on the probability"},{"startTime":3258.63,"endTime":3261.0,"body":"of experiencing an event."},{"startTime":3261.0,"endTime":3265.53,"body":"QALYs and costs are then estimated."},{"startTime":3265.53,"endTime":3267.72,"body":"And then in if it's assumed"},{"startTime":3267.72,"endTime":3269.658,"body":"that patients then move"},{"startTime":3267.72,"endTime":3269.658,"body":"into a Markov model,"},{"startTime":3269.658,"endTime":3273.09,"body":"so we've got a three state"},{"startTime":3269.658,"endTime":3273.09,"body":"Markov model here as an example,"},{"startTime":3273.09,"endTime":3277.23,"body":"patients can then, in the"},{"startTime":3273.09,"endTime":3277.23,"body":"model, join at time equal zero,"},{"startTime":3277.23,"endTime":3279.735,"body":"but they would join with the costs"},{"startTime":3279.735,"endTime":3282.81,"body":"and QALYs payoffs being"},{"startTime":3279.735,"endTime":3282.81,"body":"incurred in that first period."},{"startTime":3282.81,"endTime":3284.452,"body":"The distribution of patients"},{"startTime":3284.452,"endTime":3287.432,"body":"across the different states"},{"startTime":3284.452,"endTime":3287.432,"body":"would also be reflective"},{"startTime":3287.432,"endTime":3290.31,"body":"of the decision tree outcomes"},{"startTime":3290.31,"endTime":3292.32,"body":"and the response status they've received."},{"startTime":3292.32,"endTime":3295.86,"body":"So for example, if we have"},{"startTime":3292.32,"endTime":3295.86,"body":"different response statuses,"},{"startTime":3295.86,"endTime":3298.95,"body":"you may have say multiple"},{"startTime":3295.86,"endTime":3298.95,"body":"engines in the model"},{"startTime":3298.95,"endTime":3301.59,"body":"and the likelihood of a patient"},{"startTime":3298.95,"endTime":3301.59,"body":"being in a progressed state"},{"startTime":3301.59,"endTime":3304.17,"body":"from time equals zero in the Markov model,"},{"startTime":3304.17,"endTime":3307.692,"body":"but in the whole model process"},{"startTime":3304.17,"endTime":3307.692,"body":"it'll be time equals one year"},{"startTime":3307.692,"endTime":3310.68,"body":"could be greater than the"},{"startTime":3307.692,"endTime":3310.68,"body":"likelihood of that happening."},{"startTime":3310.68,"endTime":3314.571,"body":"If you have patients who"},{"startTime":3310.68,"endTime":3314.571,"body":"have responded to treatment,"},{"startTime":3314.571,"endTime":3317.97,"body":"in which case you would"},{"startTime":3314.571,"endTime":3317.97,"body":"have different numbers"},{"startTime":3317.97,"endTime":3320.456,"body":"of patients in the specific states"},{"startTime":3320.456,"endTime":3322.47,"body":"at the beginning of the Markov approach"},{"startTime":3322.47,"endTime":3324.96,"body":"and they'll then be"},{"startTime":3322.47,"endTime":3324.96,"body":"accruing at different costs"},{"startTime":3324.96,"endTime":3327.57,"body":"and QALYs in terms of the total amounts"},{"startTime":3327.57,"endTime":3330.355,"body":"that are accrued at each model cycle"},{"startTime":3330.355,"endTime":3334.503,"body":"compared to other response"},{"startTime":3330.355,"endTime":3334.503,"body":"statuses that have occurred."},{"startTime":3336.69,"endTime":3341.52,"body":"Okay, I think Hannah's now"},{"startTime":3336.69,"endTime":3341.52,"body":"going to discuss the taxonomies,"},{"startTime":3341.52,"endTime":3343.95,"body":"the likelihood or the key considerations"},{"startTime":3343.95,"endTime":3345.813,"body":"when selecting a model to choose."},{"startTime":3347.97,"endTime":3348.92,"body":"- Yeah. Thank you."},{"startTime":3348.92,"endTime":3350.97,"body":"So the final thing we"},{"startTime":3348.92,"endTime":3350.97,"body":"just want to cover here"},{"startTime":3350.97,"endTime":3354.15,"body":"is like the question, how do"},{"startTime":3350.97,"endTime":3354.15,"body":"you choose a suitable model"},{"startTime":3354.15,"endTime":3356.16,"body":"structure for your decision problem?"},{"startTime":3356.16,"endTime":3357.78,"body":"And there will be numerous factors"},{"startTime":3357.78,"endTime":3361.051,"body":"that guide the decision"},{"startTime":3357.78,"endTime":3361.051,"body":"on which structure to use."},{"startTime":3361.051,"endTime":3363.39,"body":"One example of a reference here"},{"startTime":3363.39,"endTime":3367.62,"body":"by Brennan et al, 2006 is the"},{"startTime":3363.39,"endTime":3367.62,"body":"taxonomy of model structures"},{"startTime":3367.62,"endTime":3369.956,"body":"for economic evaluation"},{"startTime":3367.62,"endTime":3369.956,"body":"of health technologies"},{"startTime":3369.956,"endTime":3371.1,"body":"that you can see here on the right."},{"startTime":3371.1,"endTime":3374.85,"body":"And there's some useful"},{"startTime":3371.1,"endTime":3374.85,"body":"considerations in this reference"},{"startTime":3374.85,"endTime":3377.973,"body":"if you do want to go in and"},{"startTime":3374.85,"endTime":3377.973,"body":"have a look at that separately."},{"startTime":3377.973,"endTime":3379.83,"body":"However, it's important to note"},{"startTime":3379.83,"endTime":3381.54,"body":"that it's not exhaustive"},{"startTime":3381.54,"endTime":3383.79,"body":"and there are some commonly"},{"startTime":3381.54,"endTime":3383.79,"body":"used model approaches"},{"startTime":3383.79,"endTime":3385.29,"body":"that aren't mentioned here."},{"startTime":3385.29,"endTime":3387.513,"body":"For example, partitioned"},{"startTime":3385.29,"endTime":3387.513,"body":"survival modelling."},{"startTime":3389.16,"endTime":3392.52,"body":"So we just wanted to"},{"startTime":3389.16,"endTime":3392.52,"body":"finish off with a checklist"},{"startTime":3392.52,"endTime":3395.739,"body":"of 10 things to ask when"},{"startTime":3392.52,"endTime":3395.739,"body":"selecting a model structure."},{"startTime":3395.739,"endTime":3399.87,"body":"This choice is usually made"},{"startTime":3395.739,"endTime":3399.87,"body":"relatively early in projects,"},{"startTime":3399.87,"endTime":3402.96,"body":"so it is often not possible"},{"startTime":3399.87,"endTime":3402.96,"body":"for model developers to switch"},{"startTime":3402.96,"endTime":3404.52,"body":"to a different model structure"},{"startTime":3404.52,"endTime":3405.99,"body":"and therefore it's really important"},{"startTime":3405.99,"endTime":3407.76,"body":"to thoroughly consider the options"},{"startTime":3407.76,"endTime":3410.973,"body":"and choose an appropriate"},{"startTime":3407.76,"endTime":3410.973,"body":"model structure early on."},{"startTime":3412.009,"endTime":3413.25,"body":"So first of all,"},{"startTime":3413.25,"endTime":3416.73,"body":"which structure really"},{"startTime":3413.25,"endTime":3416.73,"body":"represents the disease pathway?"},{"startTime":3416.73,"endTime":3419.037,"body":"This is something that we've"},{"startTime":3416.73,"endTime":3419.037,"body":"talked about throughout I think"},{"startTime":3419.037,"endTime":3420.912,"body":"and is one of the most or the first things"},{"startTime":3420.912,"endTime":3422.924,"body":"that should be considered."},{"startTime":3422.924,"endTime":3425.52,"body":"One way to help with this is looking"},{"startTime":3425.52,"endTime":3427.2,"body":"whether there's any"},{"startTime":3425.52,"endTime":3427.2,"body":"precedent for modelling"},{"startTime":3427.2,"endTime":3428.683,"body":"in the disease area."},{"startTime":3428.683,"endTime":3431.52,"body":"And then number three is what quantity"},{"startTime":3431.52,"endTime":3432.75,"body":"of data will be available"},{"startTime":3432.75,"endTime":3435.06,"body":"and how long is the follow-up duration?"},{"startTime":3435.06,"endTime":3436.798,"body":"And specifically does this data,"},{"startTime":3436.798,"endTime":3439.89,"body":"does the data available"},{"startTime":3436.798,"endTime":3439.89,"body":"particularly lend itself"},{"startTime":3439.89,"endTime":3441.27,"body":"to a certain model structure?"},{"startTime":3441.27,"endTime":3444.399,"body":"So for example, do you have"},{"startTime":3441.27,"endTime":3444.399,"body":"patient-level data available"},{"startTime":3444.399,"endTime":3447.449,"body":"to be able to do a"},{"startTime":3444.399,"endTime":3447.449,"body":"discrete event simulation?"},{"startTime":3447.449,"endTime":3449.16,"body":"If you have overall survival"},{"startTime":3449.16,"endTime":3451.05,"body":"and progression-free survival data"},{"startTime":3451.05,"endTime":3453.27,"body":"for a partitioned survival model"},{"startTime":3453.27,"endTime":3456.24,"body":"or transition probabilities"},{"startTime":3453.27,"endTime":3456.24,"body":"for a Markov model?"},{"startTime":3456.24,"endTime":3459.21,"body":"Will the population experience"},{"startTime":3456.24,"endTime":3459.21,"body":"events in discrete time?"},{"startTime":3459.21,"endTime":3460.38,"body":"In which case you may want"},{"startTime":3460.38,"endTime":3463.875,"body":"to look at a discrete event"},{"startTime":3460.38,"endTime":3463.875,"body":"simulation or a Cox regression"},{"startTime":3463.875,"endTime":3466.331,"body":"or in continuous time periods,"},{"startTime":3466.331,"endTime":3468.48,"body":"in which case you may"},{"startTime":3466.331,"endTime":3468.48,"body":"want to look at something"},{"startTime":3468.48,"endTime":3469.503,"body":"like a Markov model."},{"startTime":3470.777,"endTime":3473.01,"body":"Whether there's a large"},{"startTime":3470.777,"endTime":3473.01,"body":"amount of heterogeneity"},{"startTime":3473.01,"endTime":3475.92,"body":"in the patient population will help inform"},{"startTime":3475.92,"endTime":3478.055,"body":"whether to use a patient-level approach."},{"startTime":3478.055,"endTime":3481.08,"body":"The time horizon is also"},{"startTime":3478.055,"endTime":3481.08,"body":"important to consider"},{"startTime":3481.08,"endTime":3483.003,"body":"as a short time horizon may be more,"},{"startTime":3484.02,"endTime":3488.19,"body":"a decision tree would be maybe"},{"startTime":3484.02,"endTime":3488.19,"body":"more suitable in that case."},{"startTime":3489.156,"endTime":3491.37,"body":"Do the necessary assumptions hold?"},{"startTime":3491.37,"endTime":3493.61,"body":"So if you are looking at a Markov model,"},{"startTime":3493.61,"endTime":3496.464,"body":"the memoryless property"},{"startTime":3493.61,"endTime":3496.464,"body":"needs to hold, for example."},{"startTime":3496.464,"endTime":3499.77,"body":"Do risks and transitions change over time?"},{"startTime":3499.77,"endTime":3504.09,"body":"If so, you may need to"},{"startTime":3499.77,"endTime":3504.09,"body":"include a semi-Markov approach"},{"startTime":3504.09,"endTime":3506.55,"body":"or look to do more complex modelling"},{"startTime":3506.55,"endTime":3509.07,"body":"such as a discrete event simulation."},{"startTime":3509.07,"endTime":3511.861,"body":"And the last point is more"},{"startTime":3509.07,"endTime":3511.861,"body":"of a practical consideration"},{"startTime":3511.861,"endTime":3514.65,"body":"whether a simple approach is needed"},{"startTime":3514.65,"endTime":3518.883,"body":"due to external factors such"},{"startTime":3514.65,"endTime":3518.883,"body":"as time and data restraints."},{"startTime":3520.716,"endTime":3524.613,"body":"So we did receive lots of questions"},{"startTime":3524.613,"endTime":3528.276,"body":"and thank you very much"},{"startTime":3524.613,"endTime":3528.276,"body":"for submitting these."},{"startTime":3528.276,"endTime":3530.82,"body":"Because we've run out of time,"},{"startTime":3530.82,"endTime":3533.76,"body":"I think we'll probably"},{"startTime":3530.82,"endTime":3533.76,"body":"look to answer those,"},{"startTime":3533.76,"endTime":3535.92,"body":"follow up with those"},{"startTime":3533.76,"endTime":3535.92,"body":"individually afterwards"},{"startTime":3535.92,"endTime":3539.4,"body":"rather than covering any"},{"startTime":3535.92,"endTime":3539.4,"body":"questions at this point."},{"startTime":3539.4,"endTime":3541.86,"body":"So yeah, thank you very much"},{"startTime":3541.86,"endTime":3543.63,"body":"and if you have any"},{"startTime":3541.86,"endTime":3543.63,"body":"additional ones to add,"},{"startTime":3543.63,"endTime":3544.68,"body":"please go ahead and do that"},{"startTime":3544.68,"endTime":3546.98,"body":"and we will try our best"},{"startTime":3544.68,"endTime":3546.98,"body":"to reply to them all."},{"startTime":3550.29,"endTime":3552.36,"body":"And finally,"},{"startTime":3552.36,"endTime":3555.69,"body":"and in case you're wondering"},{"startTime":3552.36,"endTime":3555.69,"body":"how you can get in touch"},{"startTime":3555.69,"endTime":3557.88,"body":"with us to be able to support you,"},{"startTime":3557.88,"endTime":3559.26,"body":"we do offer introductory"},{"startTime":3559.26,"endTime":3561.18,"body":"and intermediate level training sessions"},{"startTime":3561.18,"endTime":3563.315,"body":"on key modelling concepts."},{"startTime":3563.315,"endTime":3566.22,"body":"Here on the left, you"},{"startTime":3563.315,"endTime":3566.22,"body":"can see some examples"},{"startTime":3566.22,"endTime":3568.751,"body":"of training sessions"},{"startTime":3566.22,"endTime":3568.751,"body":"that we have available."},{"startTime":3568.751,"endTime":3571.74,"body":"And of course, please"},{"startTime":3568.751,"endTime":3571.74,"body":"do reach out to our team"},{"startTime":3571.74,"endTime":3572.7,"body":"if you want support"},{"startTime":3572.7,"endTime":3575.88,"body":"with developing your next"},{"startTime":3572.7,"endTime":3575.88,"body":"health economic model."},{"startTime":3575.88,"endTime":3579.133,"body":"We have a wide range of models"},{"startTime":3575.88,"endTime":3579.133,"body":"that we can support with"},{"startTime":3579.133,"endTime":3582.6,"body":"and we'd be happy to get in touch with you"},{"startTime":3582.6,"endTime":3584.49,"body":"here at the Mtech Health Economics team."},{"startTime":3584.49,"endTime":3587.07,"body":"So you can email us below for next steps"},{"startTime":3587.07,"endTime":3589.56,"body":"or if you want to put a"},{"startTime":3587.07,"endTime":3589.56,"body":"message in the chat now,"},{"startTime":3589.56,"endTime":3593.19,"body":"then we can contact you after the webinar."},{"startTime":3593.19,"endTime":3596.46,"body":"We'd be really pleased to"},{"startTime":3593.19,"endTime":3596.46,"body":"hear how you found the webinar"},{"startTime":3596.46,"endTime":3598.17,"body":"and to get in touch."},{"startTime":3598.17,"endTime":3601.83,"body":"So thank you to Danny as"},{"startTime":3598.17,"endTime":3601.83,"body":"well for the presentation"},{"startTime":3601.83,"endTime":3604.62,"body":"and I think that's everything."},{"startTime":3604.62,"endTime":3606.858,"body":"So, thank you again."},{"startTime":3606.858,"endTime":3607.783,"body":"- Thank you very much, everyone."},{"startTime":3610.08,"endTime":3611.67,"body":"- [Announcer] Thank you for watching."},{"startTime":3611.67,"endTime":3613.92,"body":"If you'd like to find"},{"startTime":3611.67,"endTime":3613.92,"body":"out more about our work"},{"startTime":3613.92,"endTime":3616.83,"body":"or how we could support"},{"startTime":3613.92,"endTime":3616.83,"body":"your market access goals,"},{"startTime":3616.83,"endTime":3621.75,"body":"please email info@mtechaccess.co.uk"},{"startTime":3621.75,"endTime":3625.923,"body":"or visit our website at mtechaccess.co.uk."}]}