Michael Braun-Boghos

Embark on a journey through the dynamic world of pharmacovigilance with Michael Braun-Boghos, the seasoned Senior Director of Safety Strategy at Oracle. In our first  episode of the season, Michael unpacks his three-decade odyssey from the analog days of adverse event forms to the digital frontier of pharmacovigilance. As we traverse his career landscape, the evolution from paper trails to strategic IT roles unfolds, highlighting pivotal shifts and the integrative prowess of systems like Argus. Michael's anecdotes from the merge that birthed Astellas to the strategic tides at Relsys before its acquisition by Oracle offer a rare glimpse into the industry's transformation.

This episode is a deep dive into the sea of innovation that is healthtech, where the currents of patient safety and regulatory compliance meet. Michael defends the gradual innovation ethos that Argus embodies within the high-stakes realm of pharmacovigilance, underscoring its necessity amidst an industry where risk is a constant companion. Our discussion also sails into the partnership between software providers and health authorities, the burgeoning concept of 'touchless cases', and the cautious yet crucial adoption of AI. Prepare to be enlightened by our collaborative narrative that stresses the importance of evolving procedures to harness technology's full potential without sacrificing safety.

Looking towards the horizon, we cast a net over the transformative impact of automation and AI on pharmacovigilance. Michael paints a future where Argus and similar tools streamline processes, and the integration of real-world data is set to revolutionize drug safety monitoring. We tackle the utilization of diverse data sets, from electronic health records to the untapped vistas of social media, and discuss how this influx of information is carving a path for precision in pharmacovigilance. Michael's personal reflections on the intrinsic rewards of contributing to global health safety, coupled with an exploration of the precision needed for enhanced patient outcomes, brings our episode to a close, leaving listeners with a profound understanding of the commitment to innovation and patient safety that defines the field's future.

Transcript

Indy Ahluwalia 0:19

Hi and welcome to season two of our podcast Innovation and the Future of Pharma Covidence. This is a podcast series brought to you by Trudiant Talks. I'm your host, indy Alawalia, and I'm delighted to navigate the dynamic world of pharma covidence and risk management with you. As always. A quick disclaimer the opinions expressed in this episode are solely those of the individual guest and do not necessarily reflect the official views of Trudiant Consulting or their own company. We're all about fostering insightful conversations here at Trudiant Talks and we want you to know that any product, vendor or service mentioned does not imply an endorsement. If you're seeking any professional advice for specific situations, we encourage you to go speak to our experts. Please remember this podcast content is meant for informational and educational purposes only. So season two and for episode one, we have today the incredible fortune to have the magnificent MBB Senior Director Safety Strategy at Oracle as our guest speaker. Michael, thank you so much for coming in and being our opener for season two.

MBB 1:41

Thanks, Indy, for inviting me. I'm really excited to talk to you about innovation. It's one of my favorite topics.

Indy Ahluwalia 1:50

And Michael, you've been around the circuit, as it were, of PV, but what I really want to know is how did you get started in PV?

MBB 2:01

Oh well, that was many, many years ago, actually about 30 years ago now. I started in the PV department of Fujisawa. I was a student at that time and my first job ever in PV was filing AE forms into patient binders. At that time everything was still paper, right. So we used to get the AE forms on paper with carbon copy real carbon copy and I was pulling those apart and filing those in different binders and then I got offered a full-time job there. I started out doing named patient use, compassionate use, so kind of administering that program. Then I did data entry of adverse events and then I really did almost everything in the PV department, learning by doing right. So I studied linguistics, so nothing to do with pharma. I just happened to get that job filing papers because somebody I knew in the pharmacovigilance department told me they have a student job open.

Indy Ahluwalia 3:22

I think that's the way a lot of people that they just suddenly oh, here we are, we're in PV, yes yeah.

MBB 3:32

So then I started you know, they kind of saw that I had some talent on more on the database.

MBB 3:41

I started doing more IT type things.

MBB 3:46

I was actually in the PV department, but kind of the interface between the business side of PV and then the actual IT department, and then in the end I was managing a small group three or four people called the safety data and quality management team. So the safety data management was basically being an admin for Argus, a business admin, so not a DBA, but like going into, you know, making the configurations in Argus for studies and products and things like that and also doing data retrieval, going in and running reports and for people in the department. And then the safety quality management side was writing SOPs for the IT department. So yeah, that was a really interesting mix and got to do a lot of exciting things. We, you know we used to have a homegrown system when I started out there, so I would, when we needed a new feature, I would just go down to the IT department, you know one level lower, and say, oh, could you put a field like this on the screen here? And they would like do it in real time and like how about this.

MBB 5:11

Do you like this? This was the day before validation and all that. And then, yeah, the company decided to get Argus and migrate from our homegrown system, so it was involved in that project and then, you know, argus upgrades and then, of course, e2b came along right, where we moved from paper to electronic, and that was that was a really big deal. Right, the EMA was there, but they didn't really have a lot of the authority that they have now, so we had to, you know, test E2B with each authority in Europe and, yeah, that was quite a long but very interesting project.

MBB 5:59

So, yeah, and I did that for like 13 years, fujisawa eventually became Estelas, yeah, so I was there in the PV department for 13 years. Then eventually I went to the vendor of Argus, which was Relsys, and started in the strategy team there, and then after two years we were acquired by Oracle. That's how I came to Oracle, stayed in the strategy team and now I've been at Oracle for 15 years. So, yeah, it's the time flies by, but it's actually been like three decades now doing PV. Wow.

Indy Ahluwalia 6:44

Wow, and you like to stay places as well. You like to stay places for a long time.

MBB 6:52

I do, I admit it.

Indy Ahluwalia 6:58

And you've had two very interesting changes in your career. Then you went from Fujisawa and then did it get acquired by Estelas, or did it eventually just become Estelas?

MBB 7:10

It was a merger, really, between two similarly sized Japanese companies, so my company was Fujisawa, the other company was Yamenuchi. And then they merged and changed the name to Estelas, so that was interesting. Also, working for a Japanese company was very interesting for 13 years and I'm originally American but living in Germany for 35 years now, so I was American living in Germany, working for Japanese companies Very global experience, I guess.

Indy Ahluwalia 7:54

Yeah, and what was that experience like? I mean, obviously, compared to staying at Fujisawa, you didn't spend so much time at the combined Estelas. Was there integration problems at the time, or was it just the case that you felt it was time to move on?

MBB 8:17

So actually there was a program to look at the two safety systems. So Yamanuchi and Kujisawa had two different safety systems. We looked at both ones and we also looked at other safety systems and decided in the end to move everybody onto ARCIS. So there was a migration project which I was part of and that was obviously successful. And then the reason for them for leaving at the end was actually that they had from the two predecessor companies, fujisawa and Yamanuchi.

MBB 9:03

They had two European headquarters, one in Munich where I live and one in the Netherlands, and then after some years of doing that and it was separated by products basically. So the Munich Center took care of the products from Fujisawa, legacy Fujisawa, and vice versa, and then they decided to merge the two departments into one and they decided to close the Munich office. And so that was the point when I and a lot of other people who wanted to stay in Munich had a house and family in Munich. That was kind of the decision.

Indy Ahluwalia 9:58

Ah, so pretty straightforward. And then the second thing major that happened in your career was, obviously you went to RELSIS and then RELSIS were acquired by Oracle. I mean, that must have been quite a different state, from a small company into a large conglomerate company.

Innovation and Challenges in Safety Software

MBB 10:17

Yeah, that was also quite interesting. You know, I learned about RELSIS through just being a customer, through being an Argus customer, and then I started going to. They had annual user group meetings and I would often go to those and those were really interesting. You got to meet the people at RELSIS, got to meet the other customers often, was able to do presentations on what we were doing with Argus and hear how other customers are using Argus. So that's how I kind of got to know the RELSIS team and really liked them and that was kind of a big part of the decision to go to RELSIS. Yeah, with the acquisition then. So it's quite interesting. Relsis was a pretty small company right In Europe. I think we were five people altogether.

Indy Ahluwalia 11:19

Wow, I didn't realize it was that small.

MBB 11:21

Yeah, I mean, the whole company was bigger than that. I think we were probably under 30 people, something like that, but only five in Europe. And so in a small company like that, you end up wearing many hats, right. You jump to whatever needs to be done, right? So, although I was officially in the strategy department, the strategy was just part of my job, right. So I was involved in a fairly big project to implement, argus at a large pharmaceutical company, and just sort of jumped around to many different things that needed to be done, right. That's the way a small company works.

MBB 12:12

So with the acquisition by Oracle, I really got a chance to focus on the strategy part and because obviously Oracle has enough people to do all the different jobs and then. So that was actually quite nice in a way, to be able to focus on strategy and not have to sort of jump into a hundred other things. But of course, it's a large company and it's large company culture is different to a small company culture, and so, yeah, that is a change. But Oracle is acquiring companies all the time, so they have actually a very sort of time honored process, the way that they do that and the onboarding and so on. They're used to that. So often, when you meet other people from Oracle, first question is native or acquired? Ha, ha, ha, wow Ha.

MBB 13:26

That's fantastic, and you've been at Oracle for 15 years, yeah good little thing, 15 years now with the change for the new year, yeah.

Indy Ahluwalia 13:37

Wow, and 15 years. There Argus was. It must have been a fairly early number, maybe three or four when you first started, Is that right?

MBB 13:50

So Argus was born in 1997, actually, and went through many, many innovations over the years. So let's see, when I joined, I think there was already the web version of Argus. Right, it started out as a client server application, then switched to web and I think shortly after I joined, we introduced Argus Cloud for the first time. We introduced Argus Analytics for the first time. We introduced the CRO mode for Argus, with the multi-tenancy supporting multiple clients. So it was a really interesting time to join and a lot of exciting things happened along the way.

Indy Ahluwalia 14:49

Yeah, I can imagine the changes from when you joined to where it is now must be really significant. There is a level of disdain towards Argus, though, for not being necessarily innovative. What's your thoughts on that? So I have heard that as well, I think there's three reasons behind that.

MBB 15:20

So, first of all, I think there's three reasons behind that. So, first of all, I think it's not actually true. I think we are quite innovative. I mean, in the last three years alone, we have introduced a completely new UI for Empirica, which makes for has like built-in dashboards, built-in visualizations, makes work much more efficient, much less clicks to get to where you wanna go and seeing basically important things on the screen and in color that you need to attend to right away. We introduced Safety One Intake, which is our very first AI-powered product in Oracle Life Sciences. So that is using AI to automatically extract information from safety source documents, whether they're structured or unstructured documents, and put them directly into Argus, which basically bypasses the manual data entry step in Argus. Right, so it's a huge time saver. What else? We just introduced our first AI feature in Argus Probably many people don't know about that and that is a feature for automatic translation of its outbound translation. So, for instance, if you have to send the narrative to a certain authority in a certain language, you can use that. So I think we are actually quite innovative, but people might not know about it. And that leads me sort of to the second reason is I think people are getting a lot of their information from third parties. Perhaps I heard that Argus hasn't changed in years that kind of thing, so I would encourage people to come to the source. Please contact me or anyone at Oracle, and we can really tell you about the innovative things that we've been doing. One good way to stay up to date is the Oracle Safety Consortium, where we meet on a monthly basis remotely, and we try to meet twice a year face to face, once in Europe, once in North America, and there you can really get very up to date news about what we're doing, also what other customers are doing, and so on.

MBB 18:07

And then the third reason, which might be the most important reason, is our approach to innovation, I think, which is called continuous innovation, which is sort of an approach to continuously innovate in an incremental way, always adding on new innovations over time, and that is in contrast to discontinuous innovation, which is much more disruptive and flashy. So I think part of it is that we have this continuous innovation approach where we are steadily innovating, bringing out new innovations with every release, but it might not be as flashy as some other vendors. However, I think that the continuous innovation approach is the right way to go in safety, because this is a high-risk business. There are patient lives at stake here, and nothing against being disruptive, which can be good. But you have to be careful in an area of high-risk like patient safety and that's why we kind of decided not to, you know, not to throw out Argus and Empirica and start over. Right, both those products were. They just both happened to be. First release was in 1997, so they're over 25 years now and instead of seeing that as a liability, we see that as an asset.

MBB 20:06

Right, because first and foremost, the safety system has to be compliant. Right, and that compliance is not easy. Right, for people on the outside of the safety world looking in, it seems sometimes easy to make a safety system right. You just make a rules engine and you're good to go right. But you know, we've seen through experience that it's a lot more than that. Right, regulations are changing throughout the world ever faster. They're on different schedules, right. So Japan has a different release schedule from EMA and FDA, and then even things which are supposed to be worldwide standards like E2BR3, every country or region has their own flavor right when they have their own fields right, and so things are constantly changing.

MBB 21:03

The health authorities now have a trend, a new trend, which is, even if they're not sort of finished with the new regulation, they already put it into production and then they kind of see how it works, and then they get feedback and then they change things right and then they're giving pharma companies ever smaller and smaller time periods to prepare. So, yeah, it's a very difficult situation to have control over and make sure that you're compliant. And you know it took us years of hard work, working with the authorities, working with customers, to get you know Argus and Empirica compliant and we don't want to throw that away. Right, because when you start from brand new it's always the problem with 1.0 software.

MBB 21:59

right, and you have to work out the kinks and there's going to be lots of bugs and you know, trying to be compliant with all these regulations around the world, as they're in movement, changing is really really hard. So, you know, we see the maturity of the systems as an asset. It's a strong foundation that we innovate on top of right. So we're adding AI features to Argus, to Empirica, to Safety One Intake, but built on that strong foundation rather than starting from scratch. And that approach, you know, is probably makes less of an impact, less of a flash in the marketplace, where it makes it look, because the products have a long history, it looks like we're not doing anything when in fact we are. And you can see that if you, you know, if you talk to me and we can show you what we're doing right.

MBB 23:04

I think I mean truly in my heart. I think Argus and Empirica are really really good pieces of software. I think they're I honestly believe that and I think they're our strongest assets.

Indy Ahluwalia 23:19

Thank you, michael. I know that was a tough question but I had to ask it and actually you touched upon something there which is really interesting, which is how vendors interact with regulatory authorities, specifically the EMA. It's noted, you know, recently we've had two regulatory agencies go with more let's call them what everyone else calls them which is next generation safety software. We've had the MHRA obviously chosen Halo, pv from InSyfe and we had the EMA choose RXLogix. Now I remember In some meetings is some information days with the EMA where they were very much unwilling to work with vendors. It seems that they Totally gone the other way and suddenly opened up. But maybe not too To yourselves, I don't know what your, what your thoughts on that is.

MBB 24:31

So we do work with health authorities. There are a few health authorities that use our software. Yeah, in the end, that is their decision right. We, we think we have a strong offering on both the case management and signal management side of things. Of course, for health authorities, the, the signaling and analysis part, is more important Than the the case processing right, which they have to do sometimes, but most of the times that is done by the pharma companies. But but we do work with with several health authorities and actually, you know, empirica was Created through basically a co-development with the FDA, so so we were used to working with health authorities and We'll continue to do so, no matter you know if, if, if, individual health authorities decide to go with other systems. But but yeah, we're happy to to work with them and we think we have a strong offering for for them as well as for, you know, pharma companies, biotechs, medical device companies, cros.

Indy Ahluwalia 25:58

So I think we have a strong presence in in all the different life science organizations and Another thing I was just thinking about was the Maybe going back to the innovation piece which is there's. There's a lot of lofty claims being told right now. One of the ones that I hear a lot is about touchless cases, and the reality is Now there is no single software that could truly do touchless cases. In your opinion, do you think truly touchless is close or do you think PV is not ready yet but true touchless cases?

MBB 26:46

I Don't think we're there quite yet. We're on the pathway to get there. We are working on Something called augmented AI, where it's basically the AI and people working together. So in some cases the AI can make decisions. In other cases you want the AI to Sort of make an analysis and provide Candidates, let's say, but then the human makes the final decision. So let's Take duplicate check as an example. So the AI can look in the database and see if there are good candidates for for being a duplicate. It can give a score to each candidate as to how confident it is that that it is a true duplicate case. So let's say, from zero to a hundred, and it can present, you know, the user with, let's say, the top ten and with their scores, but then the user can make the final decision right. So I think we're first started sort of moving into that territory with augmented AI and, you know, until people become more familiar and, through testing, see that the AI is actually doing doing its job, before we we move to true touchless.

MBB 28:27

The other thing is, if I can sort of continue with this Continuous innovation approach I was talking about before. I think it's a mistake to try to Make your entire workflow touchless in one fell swoop. Right, because, first of all, it's a huge task. There's lots of things that you have to do. Second of all, if something goes wrong Right, and something probably will go wrong in the first attempt you won't know where the problem is right, because you've, instead of just changing one thing, you've changed a hundred things and now something went wrong along the way. And where. Where's the problem right?

MBB 29:15

So I Like to advocate this continuous innovation approach where you start with one one thing you want to automate with AI, and which thing do you start with? Well, we also have that question and we just start. We decided to start with case intake. Why? Because it's one of the steps, because of the manual data entry part of intake. That's one of the steps that takes the most time right in the work, in the workflow, and, of course, you're trying to process the cases as quickly as possible so that you get the expedited report out the door as quickly as possible, so that you stay compliant, right, and so you know. When we looked at how much time it takes in the different parts of the workflow, case intake seemed like a good return on investment, like you could focus on that and, you know, maybe get you know, reduce the time that it takes to process a case by 50%, let's say.

Indy Ahluwalia 30:23

And because because manual data takes so long right.

MBB 30:26

Mm-hmm. And then the other good thing about intake. So you can also say, for instance, medical review takes a long time, right, but with with case intake, you have, you have the correct answer there, right you have. You have the source documents that you can always look at and see did the AI do it right? Did it so? In our, in our case, safety one intake is extracting the information out of the, out of the document automatically and populating that. That extracted information into the argus fields, right, but if you, if you want to check if it did it right, you can always open the source document, which is also stored in Argus, and and just check With medical review. That's not so easy, right? So mm-hmm.

MBB 31:17

If the AI is doing the medical review and you want to check, did it do the right answer? How do you check that? And so that's why case intake seems like the ideal way to start, but you know, it might be different in different organizations. You should look at where you know in your process, where your bottlenecks are, where your pain points are, where you have a lot of manual processes, and Focus on those first and then get that right. Yeah, and you know, implement the AI or the automation if you're, if you're not using AI. And then you know, do sampling? Do a lot of sampling at the beginning.

MBB 31:58

Sampling is when you just do the process manually to check that the AI is doing it correctly, and then over time, you can reduce the sampling rate. You should probably never reduce it to zero because Inspector from the health authority can come and ask you. You know, how, are you sure that the AI is doing the right thing? But you know, even with human Humans doing the work, your, you're also doing Probably a QA or QC check every month. Right, you take some random Sampling of cases from the past month and you, you double-check them that they've been entered the right way by the humans. So it's the same thing with the AI right.

MBB 32:41

You'll do a random sampling every month and and check some of the cases and see that the AI did did the proper work. And only once you get that right then would I move into the next area and then, you know, work one area at a time Until you have a fully automated process. So that that, I think, is in a very long-winded answer to your question. That is how I think we're going to get to touchless case processing ultimately. Those saying now you know you can do touch touchless case processing today and it's easy, you know, just install our product and you move to touchless, like that I frankly I don't believe them.

Indy Ahluwalia 33:31

That's. That's a fair point. I've I like I said, I've seen a lot of lofty claims. I've seen there's another part to AI which I don't necessarily want to carry on talking about, but there are things about how, how the models will be trained, how the models will be validated. Who owns the data that trains that that model? If that model is using that data to train on, then who owns the model that? There's so many, there's so many questions that that are there, which are which are very easy to overlook but I suspect in the future Will become very prevalent in in our industry.

Indy Ahluwalia 34:19

But I was thinking about something that you'd said which was about processes, and I wanted to again go back to this question I'm sorry about about this Sort of frustration with Argus. Do you think some of it actually is due to the fact that when people Bring in a system, they could be a competitor and they move to to Argus, or they Haven't got a system and the first system that they use is Argus? It's a lot of the frustration actually from the fact that the processes have not been changed and so they expect Argus to be the process and not at all Is it. Could that be also one of the areas where it comes from I think so.

MBB 35:10

yes, we've actually been working with a lot of customers over the last five years on Optimizing their workflow and their processes. I Think this comes from the fact that many people developed their Case processing workflow in the days of paper, right? Or you literally were moving, you know, a paper, a e-form, from one person's desk to another, right, and of course things are much different today. They're very electronic. A lot of those steps in the workflow might not make any sense anymore, right, but people get sort of stuck in doing things a certain way and you do it for years and then it just becomes the way to do things right. So we've been working with quite a few customers I think it's more than 25 by now on optimizing, optimizing their workflow.

MBB 36:21

So you know, really, looking at the steps, are all those steps necessary? Do things need to be modified and are things being done manually that could be automated, right? There's actually a lot of automation features built into Argus that not everyone knows about that allow you to automate many of the steps within Argus it. You know, some of those automations may make sense for you as a company, some may not, some may depend on what kind of case it is. You know you may want to put more automation on non-serious cases than serious cases, for example, but I agree with you.

MBB 37:10

A lot of things can be helped, made more efficient, by looking at the processes and seeing how those can be made more efficient. And, of course, when you automate, when you move to AI, you also have to look at changing your processes right To adapt to that new technology. People will be doing things in a different way, sometimes slightly different, sometimes more different, and then you know you have to have people who are familiar with what the AI is doing, are able to, let's say, retrain the AI when that's needed, be able to validate the AI, be able to explain to an inspector what the AI is doing, right. So, yeah, you know, process change is a big part of this whole equation, I think and with the in quotes, digital transformation, right, that's part of all that.

Indy Ahluwalia 38:13

Yeah, it's interesting. Last year, we did a webinar talking about change, people in the change and culture, et cetera, and that's just as important as well. I think it tends to get knocked off when we're with regards to any sort of digital transformation. I think the conversation about people always tends to go away. Another thing that we've spent a lot of time in the last series of the podcast talking about is about the fact that there are less entry level roles in PV now because, yeah, they're either outsourced to big vendors in either Eastern Europe or in Asia and someone like yourself, someone like me, who actually started very junior in PV, and the roles that will probably be around. Like you said, you were talking about your augmented AI I think I've called it human in the loop in the past which is you need some experience to be able to do any of that. So where do the entry level jobs come from in your opinion?

MBB 39:37

Yeah, that is a good question. I mean, I think perhaps people are now starting more at CROs rather than the pharma company directly. It's also, of course, when people are looking at areas to study. I think now data science is a very interesting area. It's gonna become more and more relevant for PV. So if you're still studying and thinking about what are relevant areas to study, I would really recommend data science, and if you can do sort of a specialization in pharmacovigilance, even better. Right, there's very few people in the world that know both of those things.

MBB 40:30

But, yeah, I think the AI isn't really getting rid of jobs. I think it is allowing the PV department to focus their human resources on more high-level activities, right, high-value activities. So, yeah, you may not be doing data entry of cases anymore, but is that really such an exciting job which you'd be more interested in doing? Medical review or signal detection, right? Or analysis of the data, querying the database and things like that. So I think there's where the pharma company had to focus a lot of their resources on kind of these routine tasks that don't bring a lot of value but have to be done. Now they can. If they give that to the AI, then they can focus their human resources on much more interesting work, I think, and so I think there will always be entry-level work, but it just it looks completely different than that, that first job that I did of where they handed me a stack of papers like this thing and said file this. I think those days are gone and I think we're all happy that they are. It was funny I was thinking about this.

Indy Ahluwalia 42:08

It was funny. I was thinking just the other day when I was at Aids Eye. We used to have paper files as well, and I remember when the discussion happened about first going into non-paper format, basically PDFs, and not printing things out. I remember the very early conversations of that and that was scandalous at the time. And even going into eFacts, like not having a physical facts. And you're right. I mean you talk about data science. I mean, let's be honest here, you were big on the analytics side for a majority of your career in Oracle, so you do have a little soft spot for that. I'm sure you do. Yeah, but the whole point of this podcast is really to talk about the innovation and future. That's the name of the podcast, and we've talked a lot about AI today. We've talked a tiny bit about data analytics, but what is the next step for PV? Where do we go from here?

Data Use and Signaling in Pharmacovigilance

MBB 43:28

So I think the answer to that is real world data. We are used to looking at certain sort of traditional data sets for many years in PV, right, we look at databases of adverse reactions. Those can be private databases, like ARGAS. Those can be public databases like FAIRs and VAIRs and VigieBase. We're used to looking at literature databases, right Literature articles. We're used to looking at clinical trial data. Sae is coming from clinical trials.

MBB 44:16

What I think is a really exciting new area is the so-called secondary use of data. So that is real world data that was designed for some other purpose, not really for safety, but can be used for safety purposes, right? So we're talking about electronic health records, right? We're talking about administrative claims data which are basically used to decide how to fund healthcare. So that can either come from the health insurance company or it can come from the government, right? So those kinds of real world data sets I think are a treasure trove because you could find signals in there, potentially, or maybe you could validate signals that you find. So you find the signal in the traditional data sets and then you wanna validate it in the real world data. So I think that is a really interesting area and I think that, in addition to moving to AI, I think this is where the future of pharmacovigilance is going.

MBB 45:39

Now the challenge is when you have more and more data sets, how do you evaluate the answers that you're getting across? Multiple data sets, right, and we've seen that different data sets have different strengths, right? Clinical trial data is usually very good data, validated data there's just not a lot of it, right, because there's not many patients in a clinical trial. So each data set has kind of its strengths and weaknesses. So the first challenge is to kind of develop algorithms that are specific to that data set, to kind of take advantage of the strengths of that data set and get the best results.

MBB 46:32

And then, when you get results from these different data sets, let's say you're looking for a signal score for a particular product event combination. So let's say you get one signal score from data set A and a different signal score from data set B and another signal score from data set C. So which one is correct, right? So we are working on this thing which we call multimodal signaling. That kind of aggregates the signal scores across multiple data sets and comes up with one signal score that kind of takes into account the different signal score. So trying to help people understand. You know getting more data is great In safety. More data is always better than how to interpret that data is the next challenge and we're trying to work on ways to help people do that.

Indy Ahluwalia 47:37

I think that's fascinating. When you were talking about where the data comes from, even the real world data you were talking about. That's still quite sanitized data, though, isn't it? There's a biggie out there which no one wants to touch, which is actual real world social media data, which I can see why people don't want to touch, but it's still something untapped right. It's still a problem that people don't know the answer to.

MBB 48:11

That's right. We're also looking into that and some companies are doing that as well, some pharma companies. There has been, you know, a few years ago there was this PVP project, private public partnership, called WebRadar, where they looked at that exactly. Right, they did a study over several years and they looked at, you know, signals that they could find in social media data and then you know whether they would have found those signals earlier using the social media data. I think the answer was pretty much negative, right, the conclusion was that it didn't really bring a value. But, you know, it's still, I think, an open question. I think people are still looking at that.

MBB 49:10

There's, of course, different kinds of social media data, right, there's, you know, facebook and Twitter or X, I guess no. And then there's, like, really focused social media, like patients like me, right, where you have patients that might all share a particular illness are getting together and discussing among themselves and you know part of what they're discussing might be what therapies they're taking and if they're having any side effects from those therapies. Right, and you might look at that data as having a higher veracity than you know, something that's coming off of Facebook, so that, I think, is an element of it, but I think it's something we're continuing to explore and could very well be a part of this big data landscape that we're looking at, of all these different data sets. And you're right, it is a form of real world data as well.

Indy Ahluwalia 50:20

Yeah, and actually, again, it got me thinking. I was thinking of the big data companies. Obviously, Oracle is one of them, but you have AWS, you have Google, you have obviously Facebook and WhatsApp and all that whole company Meta, is it Right? So you have these big conglomerates that hold a lot of data and the syncing of that data will obviously be crucial. We're talking light years away from now, but it is still something that needs to be considered in the future about how people get the right data to be able to prescribe the correct drugs for the particular syndromes that they've got, and I had that conversation with Tony DeSousa at the end season one and I can't forget it. It's fascinating, and I guess what I'm trying to get. That is that we must remember that patient safety is the ultimate goal of what we're doing, and as long as we can get the right drugs to the right people at the right time, then we're doing our jobs correctly.

Precision Pharmacovigilance and Patient Safety

MBB 51:43

Exactly exactly, and we're actually working on a concept. So what you just described is precision medicine, right, getting the right medication to the right patient at the right time. We're working on something called precision pharmacovigilance, which actually uses this real-world data as a component, and it's about avoiding the wrong medication to the wrong patient at the wrong time or identifying those risks. Basically, right, and you're right. That is what it's ultimately all about. It's protecting the patient from harm, and I think we have to keep that in mind as we think about these new technologies and putting them in place. We have to be careful, right, and we have to be compliant, and we have to keep the patient in the center of the picture. That's the ultimate thing that we're doing. That's why I'm doing this job. Actually, I never expected to be working in pharmacovigilance, but as soon as I started, I realized, wow, I can make a real impact on the world. Right, I can make the world safer for people, which is really fantastic, right.

Indy Ahluwalia 53:15

Michael, thank you so much for joining me today. It's been a fascinating conversation and I'm glad you answered some of the more tougher questions which I wanted to ask. So thank you.