{"version":"1.0.0","segments":[{"startTime":25.0,"endTime":31.92,"body":"You know, usually when we talk about a medical diagnosis, there is this fundamental expectation of precision."},{"startTime":32.36,"endTime":33.38,"body":"It's almost like engineering."},{"startTime":33.7,"endTime":33.78,"body":"Right."},{"startTime":33.980000000000004,"endTime":40.56,"body":"Like if you fall and break your arm, you go to the hospital and the x-ray shows that jagged white line fracturing the radius."},{"startTime":41.199,"endTime":44.24,"body":"And the doctor just points at the illuminated screen and says, there it is."},{"startTime":44.620000000000005,"endTime":45.3,"body":"That is the problem."},{"startTime":45.46,"endTime":47.56,"body":"Yeah, it's a beautifully binary system."},{"startTime":48.06,"endTime":50.86,"body":"You know, you are either broken or you're not broken."},{"startTime":50.89,"endTime":52.72,"body":"You have a pathogen or you don't."},{"startTime":52.94,"endTime":56.16,"body":"And there is something deeply comforting about that, even when the news is bad."},{"startTime":56.28,"endTime":56.56,"body":"Absolutely."},{"startTime":56.86,"endTime":63.858999999999995,"body":"We desperately want our biology to be visible, to be categorized neatly into a single, you know, spreadsheet column."},{"startTime":64.0,"endTime":65.64,"body":"Because it implies a straightforward fix."},{"startTime":65.9,"endTime":66.38,"body":"Exactly."},{"startTime":66.57000000000001,"endTime":69.96,"body":"But then you step out of the emergency room and into the world of chronic illness."},{"startTime":70.66,"endTime":75.98,"body":"into the realities of aging and into the slow compounding ways our bodies change over decades."},{"startTime":76.19,"endTime":77.34,"body":"Which is where things get messy."},{"startTime":77.76,"endTime":78.539,"body":"Really messy."},{"startTime":79.14,"endTime":80.84,"body":"Suddenly that x-ray machine is useless."},{"startTime":81.16,"endTime":84.94,"body":"We are looking at a diagnostic landscape that isn't a clear photograph,"},{"startTime":85.66,"endTime":89.42,"body":"but a remarkably murky, sprawling, chaotic ecosystem."},{"startTime":89.8,"endTime":94.4,"body":"And that is incredibly frustrating for both the patient and the physician."},{"startTime":95.18,"endTime":98.98,"body":"Because chronic disease doesn't just snap into existence the way a bone breaks."},{"startTime":99.62,"endTime":100.26,"body":"It creeps."},{"startTime":100.38000000000001,"endTime":101.36,"body":"It totally creeps."},{"startTime":101.4,"endTime":107.0,"body":"So I want you to imagine your health not as a series of random isolated snapshots like"},{"startTime":107.4,"endTime":114.0,"body":"a surprise diabetes diagnosis at 50, an unexpected heart attack at 55, a sudden autoimmune flare"},{"startTime":114.0,"endTime":114.74000000000001,"body":"up at 60."},{"startTime":115.0,"endTime":115.14,"body":"Right."},{"startTime":115.36,"endTime":120.78,"body":"But instead, imagine it as a continuous, highly predictable, continuously running movie"},{"startTime":121.16000000000001,"endTime":126.12,"body":"What if the ailments you develop at age 60 were actually quietly, invisibly telegraphed"},{"startTime":126.15,"endTime":129.79999999999998,"body":"by the seemingly minor disconnected doctor's visits you had at age 30?"},{"startTime":130.1,"endTime":130.92,"body":"It's a huge what if."},{"startTime":131.07999999999998,"endTime":134.44,"body":"What if the data was always there, but we just didn't have the math to read the script?"},{"startTime":134.52,"endTime":137.38,"body":"Well, that is the exact paradigm shift we are unpacking today."},{"startTime":137.88,"endTime":141.68,"body":"We're looking at the death of reactive medicine, waiting for that jagged white line to appear,"},{"startTime":142.42,"endTime":145.5,"body":"and the birth of continuous, proactive algorithmic prediction."},{"startTime":145.68,"endTime":146.68,"body":"Okay, let's unpack this."},{"startTime":146.9,"endTime":152.4,"body":"The mission for today's Deep Dive is to explore a monumental structural shift in precision medicine."},{"startTime":152.44,"endTime":153.56,"body":"It really is monumental."},{"startTime":153.82,"endTime":163.32,"body":"We have a formidable stack of research papers today, centered primarily around a breakthrough called a Bayesian framework for longitudinal EHR and genetic discovery."},{"startTime":164.2,"endTime":178.61999999999998,"body":"We are going to explore how a multidisciplinary team of scientists is taking decades of incredibly messy longitudinal hospital data, your electronic health records, and mathematically fusing them with your underlying genetic code."},{"startTime":178.76,"endTime":182.439,"body":"To create dynamic time traveling predictions of your future health."},{"startTime":182.84,"endTime":183.17999999999998,"body":"Exactly."},{"startTime":183.54,"endTime":185.28,"body":"It really operates like a biological time machine."},{"startTime":185.61999999999998,"endTime":193.66,"body":"We are looking at computational frameworks that don't just, you know, guess your risk for next year, but actually project your entire biological trajectory decades into the future."},{"startTime":193.88,"endTime":197.06,"body":"And it dynamically updates every single time you interact with the health."},{"startTime":197.1,"endTime":197.48,"body":"care system."},{"startTime":197.82,"endTime":198.06,"body":"Precisely."},{"startTime":198.26,"endTime":199.28,"body":"So here is our roadmap."},{"startTime":199.88,"endTime":205.34,"body":"We are going to trace the historical evolution of this quest, wade into the absolute nightmare"},{"startTime":205.42,"endTime":210.7,"body":"of harmonizing raw hospital data, and then introduce this groundbreaking Bayesian model"},{"startTime":210.88,"endTime":214.0,"body":"which has the truly magical name, Alodi Nulai."},{"startTime":214.34,"endTime":215.16,"body":"Such a great name."},{"startTime":215.45999999999998,"endTime":219.98,"body":"We will look at exactly how it mathematically outmaneuvers our current gold standard clinical"},{"startTime":220.2,"endTime":220.7,"body":"risk scores."},{"startTime":221.35999999999999,"endTime":236.34,"body":"And because we cannot talk about the frontier of predictive data without talking about Silicon Valley, we will examine how big tech is deploying massive, large language models to solve this exact same problem through a completely different architectural lens."},{"startTime":236.64,"endTime":238.66,"body":"Two very different paths to the same destination."},{"startTime":238.91,"endTime":245.39999999999998,"body":"Right. And finally, we'll figure out what this all means for you, the listener, and your future relationship with the health care system."},{"startTime":245.5,"endTime":256.14,"body":"So, to truly appreciate the magnitude of what models like Aledunio are doing, we have to recognize that this is the culmination of a quest that is literally thousands of years old."},{"startTime":256.49,"endTime":260.42,"body":"The desire to predict disease trajectory isn't a modern invention at all."},{"startTime":260.54,"endTime":261.9,"body":"Right, it's not just a Silicon Valley thing."},{"startTime":262.1,"endTime":262.84000000000003,"body":"No, not at all."},{"startTime":263.17,"endTime":267.9,"body":"If we look at the evolution and trajectory of personalized and precision medicine paper in our sources..."},{"startTime":267.719,"endTime":270.7,"body":"It actually anchors this pursuit all the way back in antiquity."},{"startTime":271.32,"endTime":275.02,"body":"Around 1550 BC, you had Hippocrates and the humoral theory."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":281.08000000000004,"endTime":284.20000000000005,"body":"blood, yellow bile, black bile, and phlegm."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":284.46000000000004,"endTime":285.219,"body":"Exactly."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":285.8,"endTime":293.8,"body":"And while their biological understanding was obviously, well, flawed by today's standards, their underlying epistemology was incredibly prescient."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":293.97,"endTime":294.42,"body":"How so?"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":294.91,"endTime":302.98,"body":"Even in those ancient Greek and Egyptian treatises, there was this fundamental driving hypothesis that unique patient characteristics exist."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":303.38,"endTime":307.66,"body":"dictate both disease susceptibility and the trajectory of recovery."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":307.84000000000003,"endTime":308.5,"body":"Oh, I see."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":308.5,"endTime":312.1,"body":"They were actively trying to individualize care based on the data points they could observe."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":312.28000000000003,"endTime":316.78000000000003,"body":"The intuition was flawless, but their data resolution was, well, checking if you had too much phlegm."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":316.98,"endTime":317.20000000000005,"body":"Exactly."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":317.53000000000003,"endTime":319.48,"body":"They were operating at an incredibly low resolution."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":320.18,"endTime":325.20000000000005,"body":"Then you jump forward to the mid-20th century, 1953, the elucidation of the DNA double helix."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":325.54,"endTime":328.98,"body":"Suddenly we unlock the foundational molecular code governing heredity."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":329.32,"endTime":329.66,"body":"Huge leap."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":329.90000000000003,"endTime":330.28000000000003,"body":"Massive."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":330.88,"endTime":333.90000000000003,"body":"Fast forward to 2003, the completion of the Human Genome Project,"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":334.28000000000003,"endTime":337.1,"body":"which provided the first reference genome to map human variation."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":337.54,"endTime":342.68,"body":"And then the 1010s hit, and we enter the next generation sequencing, or NGS era."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":342.90000000000003,"endTime":343.58000000000004,"body":"Where things get cheap."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":343.90000000000003,"endTime":344.16,"body":"Right."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":344.66,"endTime":350.76,"body":"The cost of sequencing a genome plummets from billions of dollars to a few hundred bucks,"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":351.56,"endTime":357.84000000000003,"body":"making it suddenly feasible to actually integrate dense genomic data into everyday clinical research."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":358.38,"endTime":364.26,"body":"But here is the critical failure point of modern medicine and the exact problem these new papers are trying to solve."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":364.8,"endTime":369.98,"body":"Even with all that massive genomic knowledge, the way we actually practice medicine today"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":370.31,"endTime":373.52000000000004,"body":"still relies heavily on what we can call the snapshot problem."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":373.57,"endTime":375.0,"body":"The cross-sectional illusion."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":375.20000000000005,"endTime":375.54,"body":"Right."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":376.08000000000004,"endTime":380.5,"body":"Like if I go to my cardiologist tomorrow, they are going to calculate my risk using a static"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":380.72,"endTime":380.84000000000003,"body":"formula."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":381.46000000000004,"endTime":385.76,"body":"They might use the pooled cohort equation, the PCE, to figure out my 10-year cardiovascular"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":386.33000000000004,"endTime":386.44,"body":"risk."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":387.02000000000004,"endTime":390.88,"body":"Or if we were looking at oncology, they might use the JL model for breast cancer risk."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":390.899,"endTime":393.28000000000003,"body":"risk. But these models are fundamentally static."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":393.58000000000004,"endTime":395.76,"body":"They are completely anchored in the present tense."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":395.959,"endTime":396.14000000000004,"body":"Yeah."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":396.42900000000003,"endTime":401.52000000000004,"body":"The PCE takes a cross-section of you at one specific moment in time. It asks, \"What is"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":401.53000000000003,"endTime":405.02000000000004,"body":"your age today? What is your systolic blood pressure today? What is your total cholesterol"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":405.25,"endTime":407.54,"body":"today? Do you smoke today?\" And then it spits out a percentage."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":407.94,"endTime":412.22,"body":"It's like trying to predict the final score of a baseball game and exactly who is going"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":412.22,"endTime":416.899,"body":"to hit a home run in the ninth inning by only looking at a single static photograph taken"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":417.119,"endTime":417.56,"body":"during the second."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":417.6,"endTime":420.72,"body":"That is a perfect visualization of the mathematical flaw."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":420.92,"endTime":427.70000000000005,"body":"You are completely ignoring the momentum of the game, how the pitcher's mechanics are degrading over time, the weather front rolling in."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":427.90000000000003,"endTime":429.96000000000004,"body":"You miss the entire kinetic trajectory."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":430.32,"endTime":435.68,"body":"Traditional models treat diseases and your biology in strict isolation."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":436.58000000000004,"endTime":444.8,"body":"They treat your body as a series of disconnected static events rather than a continuously evolving, highly dependent ecosystem."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":444.96000000000004,"endTime":445.04,"body":"Right."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":445.1,"endTime":453.92,"body":"If you calculate your PCE at age 40 and it says you have a low risk, but then at age 42 you develop an autoimmune condition that causes systemic inflammation."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":454.14000000000004,"endTime":455.26,"body":"Your whole profile changes."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":455.76,"endTime":464.16,"body":"Completely. The PCE from age 40 is now clinically useless. It cannot update dynamically based on the introduction of a new competing variable."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":464.54,"endTime":468.84000000000003,"body":"But let me push back here because this is the part that always drives me crazy when we talk about medical tech."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":469.36,"endTime":473.06,"body":"We have had electronic health records, EHRs for decades."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":473.959,"endTime":479.399,"body":"Hospitals have been typing my blood pressure, my prescriptions, and my weird symptoms into servers since the late 90s."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":479.58000000000004,"endTime":480.959,"body":"Oh yeah. Tons of data."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":481.06,"endTime":486.14000000000004,"body":"Right. So if we possess all this longitudinal data and we know that static models like the PCE are flawed,"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":486.72,"endTime":489.44,"body":"why hasn't a computer already connected these dots for us?"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":489.58000000000004,"endTime":491.34000000000003,"body":"Why are we only having this deep dive today?"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":491.619,"endTime":494.42,"body":"That is the structural bottleneck of the entire field."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":495.019,"endTime":497.78000000000003,"body":"And the answer is found in the harsh reality of clinical data."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":498.20000000000005,"endTime":501.06,"body":"It is an absolute unmitigated disaster."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":501.28000000000003,"endTime":502.02000000000004,"body":"Disgusting mess."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":502.24,"endTime":515.58,"body":"The paper from KU Leuven on modeling time-dependent patient trajectories, along with the PEHRT framework paper, which stands for Pipeline for Harmonizing EHR Data, they outline exactly why this is so mathematically punishing."},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":515.92,"endTime":520.9399999999999,"body":"Let's get into the mud here. Why is the data so bad first? You have the problem of irregular sampling"},{"speaker":"Okay, right. The idea that human health was dictated by the balance of four humors","startTime":521.2,"endTime":525.8,"body":"You only go to the doctor when you are sick or for a random annual checkup that you might skip for three years"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":526.399,"endTime":530.56,"body":"00 a.m."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":530.56,"endTime":538.44,"body":"To provide a perfectly spaced statistically clean data point exactly there are massive gaps of years where I am perfectly healthy and"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":538.5600000000001,"endTime":543.6800000000001,"body":"Then suddenly I have three hospital visits in a single month because I caught a weird respiratory virus"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":543.6800000000001,"endTime":547.0600000000001,"body":"And standard machine learning models despise irregular intervals."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":548.0600000000001,"endTime":553.4200000000001,"body":"A standard recurrent neural network, for example, gets back a nice, evenly spaced time series."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":554.08,"endTime":558.64,"body":"When it sees a three-year gap, it struggles to understand if nothing happened, or if the"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":558.6800000000001,"endTime":559.5,"body":"data is just missing."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":559.82,"endTime":562.28,"body":"Which brings up the second problem, actual missing data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":562.3190000000001,"endTime":574.1800000000001,"body":"Yes, the missingness is rarely at random. Maybe a doctor was rushed and didn't record a secondary symptom. Or maybe you went to an out-of-network clinic while on vacation so that data never made it into your primary EHR."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":574.499,"endTime":578.52,"body":"But the third problem seems to be the most intractable one, the coding systems."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":578.939,"endTime":580.28,"body":"The heterogeneous ontologies."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":580.52,"endTime":583.74,"body":"This is where data scientists literally lose their minds."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":584.54,"endTime":592.66,"body":"Different hospitals and even different departments within the same hospital use entirely different coding systems to describe the exact same biological event."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":593.24,"endTime":593.36,"body":"Yes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":594.08,"endTime":596.1,"body":"One system might use ICD-9 codes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":596.62,"endTime":598.0,"body":"Another uses ICD-10."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":598.62,"endTime":601.1,"body":"Another uses SNUM-DCT for clinical notes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":610.3000000000001,"endTime":610.46,"body":"Yes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":601.62,"endTime":603.7,"body":"And another uses RxNorm for medications."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":604.1800000000001,"endTime":608.72,"body":"So it's like trying to read a diary written by a doctor who is sprinting down a hospital hallway,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":609.28,"endTime":612.02,"body":"constantly switching between five different languages mid-sentence,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":612.5600000000001,"endTime":616.6,"body":"occasionally using outdated slang, and leaving random pages entirely blank."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":617.24,"endTime":620.86,"body":"That is precisely what raw EHR data looks like to a computer."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":621.0500000000001,"endTime":623.86,"body":"So before you can even dream of doing any time-traveling predictions,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":624.35,"endTime":627.26,"body":"you have to build a universal semantic translator."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":627.6800000000001,"endTime":630.1,"body":"That is what the P8 pipeline is designed to achieve."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":630.439,"endTime":638.16,"body":"How does PHR actually untangle that? Because just mapping one code to another sounds like an endless manual nightmare of spreadsheets."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":638.16,"endTime":645.059,"body":"It would be, which is why PHR doesn't rely on manual mapping. It leverages representation learning and pre-trained language models."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":645.189,"endTime":645.3190000000001,"body":"Okay."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":645.809,"endTime":651.739,"body":"Instead of looking at a medical code as just a discrete, isolated identifier, it treats the codes like words in a complex language."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":652.08,"endTime":654.84,"body":"Like training an AI to understand the context of a sentence."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":655.44,"endTime":659.1800000000001,"body":"Exactly. It uses these models to generate high-dimensional embeddings."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":659.61,"endTime":664.0600000000001,"body":"It maps all this structured and unstructured data into a vector space."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":664.0600000000001,"endTime":665.26,"body":"A vector space, meaning?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":665.72,"endTime":670.6,"body":"In this mathematical case, codes that appear in similar clinical contexts are placed closer together."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":671.38,"endTime":675.24,"body":"So an outdated ICD-9 code for a specific type of hypertension"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":675.87,"endTime":680.32,"body":"and a new SNOMED note about elevated blood pressure might look totally different in a spreadsheet."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":680.96,"endTime":684.04,"body":"But in the vector space, their embeddings are nearly identical."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":684.44,"endTime":688.36,"body":"Because the model has learned the semantic relationship, it knows they mean the same"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":688.52,"endTime":691.12,"body":"thing based on the surrounding context of the patient's journey."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":691.76,"endTime":696.76,"body":"Precisely. It acts as an ontological harmonizer. It effectively takes the messy, multilingual,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":697.38,"endTime":702.44,"body":"irregularly sampled diary and translates it into a single, cohesive, machine-readable,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":702.76,"endTime":704.1800000000001,"body":"mathematically continuous timeline."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":704.54,"endTime":708.88,"body":"So they are literally building the foundational track before they can even attempt to run the predictive"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":708.9,"endTime":715.26,"body":"train. You cannot do any of the advanced multimodal Bayesian modeling we're about to discuss without that standardized track."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":716.0600000000001,"endTime":720.08,"body":"The data harmonization is the invisible heavy lifting of precision medicine."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":720.41,"endTime":724.82,"body":"Which perfectly sets the stage for the cast of characters who decided to actually build the train."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":725.0,"endTime":728.26,"body":"Our primary source today is this monumental paper,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":728.72,"endTime":732.7,"body":"\"A Bayesian Framework for Longitudinal EHR and Genetic Discovery.\""},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":733.26,"endTime":736.6800000000001,"body":"And the team behind this is a multidisciplinary powerhouse."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":736.94,"endTime":737.6,"body":"They really are."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":737.8000000000001,"endTime":741.6,"body":"You have Dr. Sarah Erbit, who is a physician scientist."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":742.0,"endTime":745.86,"body":"She is a working cardiologist, meaning she is on the front lines"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":746.22,"endTime":750.0600000000001,"body":"looking at these static PCE scores and watching them fail real human beings."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":750.34,"endTime":750.52,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":750.7,"endTime":754.0600000000001,"body":"But she also has incredibly deep computational expertise."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":754.619,"endTime":764.999,"body":"She teamed up with Alexander Gusev, Pradeep Natarajan, and Giovanni Parmigiani, drawing resources across institutions like Mass General Brigham, Harvard, and the Broad Institute."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":765.34,"endTime":775.66,"body":"This is a vital point about the sociology of science. This isn't just a group of computer scientists playing with health data in a vacuum, and it isn't just a group of doctors writing opinion pieces about algorithms."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":776.1,"endTime":776.259,"body":"Yeah."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":776.48,"endTime":785.4200000000001,"body":"This is a team that intimately understands both the brutal, messy reality of clinical pathophysiology and the rigorous mathematics required to model it without overfitting."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":785.739,"endTime":793.78,"body":"And when this team looked at the landscape of predictive medicine, they saw a field that was repeatedly slamming into a series of deeply frustrating dead ends."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":794.3000000000001,"endTime":798.98,"body":"Understanding these dead ends is crucial because it explains why their new approach is so radical."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":799.36,"endTime":800.22,"body":"It sets the baseline."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":800.44,"endTime":801.2,"body":"Let's walk through them."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":801.52,"endTime":807.02,"body":"Dead end number one, which we briefly touched on, is treating diseases as isolated silos."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":807.6800000000001,"endTime":814.58,"body":"Right. If you use a traditional Cox proportional hazards model, which is a very standard statistical tool in medical research,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":815.44,"endTime":822.14,"body":"it might look at your risk of developing heart disease and then completely separately look at your risk of developing chronic kidney disease."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":823.1,"endTime":827.52,"body":"But biology is not compartmentalized. Conditions co-evolve and compound."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":827.52,"endTime":828.14,"body":"They cascade."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":828.5,"endTime":834.34,"body":"Exactly. A metabolic issue, like insulin resistance, directly drives an inflammatory cascade,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":835.0,"endTime":839.9200000000001,"body":"which damages the vascular endothelium, which leads to cardiovascular disease, which stresses the kidneys."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":839.9200000000001,"endTime":840.88,"body":"It's a domino effect."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":841.26,"endTime":845.98,"body":"If your statistical model does not natively capture the dependent interplay between these conditions,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":846.82,"endTime":851.02,"body":"If it assumes they are independent variables, it is mathematically guaranteed to miss the plot."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":851.44,"endTime":852.9,"body":"Which leads to dead end number two."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":853.96,"endTime":861.58,"body":"Some researchers realized this, so they tried to group diseases together using something called unsupervised clustering, or topic modeling."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":862.24,"endTime":864.54,"body":"Topic modeling is borrowed from natural language processing."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":865.6,"endTime":874.28,"body":"Imagine you feed an algorithm a million news articles, and it automatically groups them into sports, politics, and finance based on word frequencies."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":874.9,"endTime":877.1800000000001,"body":"Researchers tried doing this with EHR data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":877.95,"endTime":881.1,"body":"The algorithm looks at a massive pile of data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":880.84,"endTime":886.54,"body":"patient records and groups diseases together based on how frequently they co-occur in the same patients."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":886.96,"endTime":891.0,"body":"That sounds like a smart way to find the cascades you just mentioned. Why is it a dead end?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":891.48,"endTime":895.52,"body":"Because it is fundamentally retrospective. These models are great at explaining the past."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":895.8000000000001,"endTime":900.7,"body":"They can look at a 70-year-old patient and say, ah, yes, you belong to the metabolic cardiovascular topic cluster."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":900.84,"endTime":900.98,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":901.36,"endTime":905.02,"body":"But they assign you to that topic only after the events have already occurred."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":905.78,"endTime":910.66,"body":"They struggle immensely to predict what will happen next when you're only 45 and only have one minor symptom."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":911.08,"endTime":914.5600000000001,"body":"They don't model the dynamic tempo of how your risk changes as you age."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":914.8000000000001,"endTime":916.52,"body":"They are descriptive, not generative."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":917.14,"endTime":917.439,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":917.78,"endTime":922.74,"body":"And then there is dead end number three, which is frankly the most seductive trap in modern science right now."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":923.3190000000001,"endTime":924.98,"body":"Black box, deep learning."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":925.83,"endTime":932.9200000000001,"body":"Deep neural networks are unparalleled at finding hidden nonlinear patterns in massive high dimensional beta sets."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":933.41,"endTime":936.1,"body":"But they are notoriously stubbornly opaque."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":936.84,"endTime":938.46,"body":"Let's ground this in a clinical reality."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":938.86,"endTime":967.82,"body":"Imagine Dr. Urbut is sitting across from a patient. She looks at a standard black box deep learning model on her screen and she says the computer says you have an 85 percent chance of a massive myocardial infarction in the next five years. We need to start you on aggressive lipid lowering therapies and beta blockers immediately. Right. The patient naturally is going to be a patient."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":958.7,"endTime":959.86,"body":"going to panic and ask why."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":960.46,"endTime":961.38,"body":"What is driving this risk?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":961.54,"endTime":965.04,"body":"And if she's using a black box, her only honest answer is, I have no idea."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":965.29,"endTime":969.64,"body":"The computer multiplied millions of weights and biases in hidden layers, and it output"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":969.77,"endTime":970.5,"body":"this probability."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":970.75,"endTime":971.74,"body":"The math says so."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":971.88,"endTime":974.3000000000001,"body":"Which is clinically and perhaps legally unacceptable."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":974.96,"endTime":976.6800000000001,"body":"It is entirely unacceptable."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":977.48,"endTime":979.74,"body":"A physician requires interpretability."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":980.47,"endTime":986.3190000000001,"body":"They need to understand the underlying biological mechanism driving the risk in order to tailor the intervention."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":986.74,"endTime":987.28,"body":"That makes sense."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":987.64,"endTime":993.379,"body":"If the risk is driven by an inflammatory pathway, you treat it differently than if it's driven by a lipid pathway."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":993.78,"endTime":996.08,"body":"Deep learning models hide the pathway."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":996.48,"endTime":1001.7,"body":"Furthermore, these black box models often fail catastrophically when dealing with censored data,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1002.25,"endTime":1004.84,"body":"which is a huge issue in longitudinal health tracking."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1005.2,"endTime":1008.2,"body":"Yes, censored data is a fundamental statistical hurdle."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1008.96,"endTime":1012.88,"body":"Let's say you are tracking a cohort of patients to see who develops diabetes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1013.72,"endTime":1016.28,"body":"Patient A is currently 45 years old and does not have diabetes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1016.6,"endTime":1017.66,"body":"But the study ends today."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1018.09,"endTime":1019.619,"body":"What happens to patient A when they are 50?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1020.46,"endTime":1021.52,"body":"Or 60? We don't know."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1021.989,"endTime":1024.36,"body":"Their future is right censored from our current view."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1024.6599999999999,"endTime":1028.2,"body":"They didn't stop existing, they just aged out of our observation window."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1028.4389999999999,"endTime":1031.8999999999999,"body":"Right. And many standard machine learning models don't know how to handle that."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1032.36,"endTime":1035.72,"body":"They often either drop the patient from the training set, which loses valuable data,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1036.319,"endTime":1041.379,"body":"Or they assume that because the patient didn't get diabetes during the observation window, they never get it."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1041.78,"endTime":1043.54,"body":"Which totally skews the probabilities."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1043.879,"endTime":1050.919,"body":"Exactly. A robust generative model must mathematically account for the fact that patients simply haven't lived the rest of their lives yet."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1051.34,"endTime":1055.679,"body":"So Dr. Erbet and her multi-institutional team looked at these dead ends..."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1055.7,"endTime":1060.08,"body":"They realized they needed a model that could gracefully handle write-sensor data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1060.6390000000001,"endTime":1064.039,"body":"They needed a model that was totally transparent and interpretable,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1064.559,"endTime":1067.179,"body":"outputting biological mechanisms, not just percentages."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1067.659,"endTime":1071.419,"body":"And crucially, they needed a model that could explicitly synthesize"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1071.64,"endTime":1080.1200000000001,"body":"The static, unchanging nature of your DNA with the dynamic, constantly evolving nature of your aging body and your hospital visits."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1080.48,"endTime":1084.019,"body":"It is a phenomenally complex mathematical and biological ask."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1084.1100000000001,"endTime":1087.26,"body":"They needed to fuse a static baseline with a dynamic timeline."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1087.58,"endTime":1087.64,"body":"Yes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1087.75,"endTime":1088.7,"body":"But they actually built it."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1089.1100000000001,"endTime":1091.08,"body":"And they gave it a name that I am absolutely obsessed with."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1091.08,"endTime":1092.8990000000001,"body":"It sounds like something out of an arcane grimoire."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1093.88,"endTime":1094.6,"body":"A Lady Nuly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1094.9,"endTime":1096.14,"body":"It is quite the portmanteau."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1096.24,"endTime":1097.18,"body":"It's brilliant branding."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1097.3,"endTime":1108.5,"body":"It blends Aladdin, evoking the magic of a wish-granting genie, peering into the future, illuminating the unseen with Jacob Bernoulli, the 17th century Swiss pioneer of probability theory."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1109.38,"endTime":1110.34,"body":"Magic and math."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1110.8,"endTime":1116.3600000000001,"body":"What is truly fascinating here is how they structured the mathematics to avoid the dead ends we just discussed."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1117.3600000000001,"endTime":1132.38,"body":"A lot of newly is a Bayesian generative framework and the core architectural innovation and we need to be precise here because this is where the magic happens is that they formulated the model as a mixture of probabilities rather than a probability of a mixture."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1132.67,"endTime":1141.6,"body":"Okay, stop right there. I know our listeners are smart, but mixture of probabilities versus probability of a mixture sounds like an agonizingly subtle semantic difference."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1142.44,"endTime":1145.1390000000001,"body":"Why is this distinction the linchpin of the entire model?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1145.22,"endTime":1148.24,"body":"It changes everything about how the algorithm views a human being."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1148.44,"endTime":1149.019,"body":"Let's break it down."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1149.619,"endTime":1154.56,"body":"In traditional clustering models, the probability of a mixture approach the algorithm is fundamentally"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1154.759,"endTime":1158.06,"body":"trying to force a patient into a single, mutually exclusive bucket."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1158.56,"endTime":1160.019,"body":"It assumes competitive allocation."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1160.56,"endTime":1161.34,"body":"Let me try an analogy."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1161.68,"endTime":1165.74,"body":"It's like looking at a complex recipe, and the algorithm is rigidly programmed to say"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1165.8600000000001,"endTime":1168.94,"body":"this dish must be either a cake or are a stew."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1169.2,"endTime":1170.08,"body":"It cannot be both."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1170.6,"endTime":1177.1,"body":"It calculates the probability that you belong to cluster A versus the probability you belong to cluster B and then assigns you to the domino one."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1177.34,"endTime":1177.74,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1178.22,"endTime":1178.28,"body":"Yeah."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1178.49,"endTime":1184.22,"body":"It assumes that if you are a cardiovascular patient, that label dominates and defines your trajectory."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1184.64,"endTime":1186.16,"body":"But human health doesn't work like that."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1186.2,"endTime":1187.92,"body":"My biology isn't exclusive."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1188.84,"endTime":1189.8990000000001,"body":"Allergy Lee flips the map."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1190.31,"endTime":1194.72,"body":"By using a mixture of probabilities, it treats your health like a massive dynamic buffet."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1195.18,"endTime":1195.48,"body":"Yes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1195.619,"endTime":1198.019,"body":"It says you do not have to be just one thing."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1198.33,"endTime":1203.039,"body":"It calculates the weighted combination of different underlying signatures happening simultaneously."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1203.499,"endTime":1204.659,"body":"That is exactly it."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1204.869,"endTime":1210.18,"body":"It assumes that multiple latent biological processes are running in parallel inside you."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1210.5,"endTime":1222.94,"body":"You can have a moderate baseline of a metabolic signature, a heavy escalating helping of a cardiovascular signature, and a sudden dash of an inflammatory signature all operating at the exact same time at different weights."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1223.38,"endTime":1232.56,"body":"So my risk for a specific disease tomorrow isn't based on what bucket I'm in today. It's the weighted sum of all these different signature probabilities churning away underneath the surface."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1232.999,"endTime":1235.8990000000001,"body":"Yes. Your overall hazard function is a composite."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1236.529,"endTime":1239.96,"body":"This is how they built a model that actually reflects clinical reality,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1240.3990000000001,"endTime":1243.46,"body":"where a 65-year-old patient isn't just dealing with one primary issue,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1243.99,"endTime":1249.72,"body":"but is managing multiple simultaneous chronic conditions that interact and persist over a lifetime."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1250.18,"endTime":1253.34,"body":"To make this incredibly complex buffet work computationally,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1253.84,"endTime":1256.6,"body":"Allodinouli is built upon three distinct pillars."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1257.1390000000001,"endTime":1259.619,"body":"Pillar 1, population-level signatures."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1260.3600000000001,"endTime":1266.28,"body":"The model ingested all this longitudinal data and discovered 21 distinct replicable disease signatures."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1266.8,"endTime":1271.18,"body":"And we must emphasize that these 21 signatures are not arbitrary statistical groupings."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1271.58,"endTime":1274.02,"body":"They map to deep, recognizable biology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1274.58,"endTime":1274.98,"body":"Like what?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1275.3,"endTime":1281.14,"body":"They found a distinct metabolic signature, a neuropsychiatric signature, a malignancy signature, an inflammatory signature."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1281.6390000000001,"endTime":1286.68,"body":"They capture how hundreds of specific diseases tend to travel together in packs over the course of a human lifetime."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1286.94,"endTime":1289.6,"body":"And these signatures are alive. They are time-dependent."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1289.82,"endTime":1294.3,"body":"Yes. The model learns the natural tempo of these processes across the population."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1295.019,"endTime":1298.619,"body":"It learns, for instance, that within the non-ischemic cardiovascular signature,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1299.18,"endTime":1305.119,"body":"the mathematical probability of developing atrial fibrillation rises steadily and specifically after age 55."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1306.279,"endTime":1313.8600000000001,"body":"or within a malignancy signature, the hazard for metastatic disease spikes sharply between the ages of 60 and 75."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1314.22,"endTime":1314.499,"body":"Wow."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1314.999,"endTime":1317.44,"body":"The model understands the chronobiology of disease."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1317.8600000000001,"endTime":1318.999,"body":"So that's the population level."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1319.3600000000001,"endTime":1322.22,"body":"Pillar 2 is where the model zooms in on you specifically."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1322.72,"endTime":1325.159,"body":"This is the concept of individual signature load."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1325.2,"endTime":1330.6390000000001,"body":"This represents how heavily you, the individual patient, map to each of those 21 signatures."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1331.019,"endTime":1334.2,"body":"And crucially, this loading is not a static number written in stone."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1334.92,"endTime":1341.98,"body":"As you age and as you unfortunately accumulate new symptoms or diagnoses in your EHR, your loading dynamically shifts."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1342.06,"endTime":1342.259,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1342.3990000000001,"endTime":1347.22,"body":"If you get diagnosed with type 2 diabetes at age 50, your loading on the metabolic signature shoots up,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1347.539,"endTime":1352.24,"body":"which automatically cascades to increase your future probability of renal failure or neuropathy."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1352.48,"endTime":1355.0,"body":"It is a continuously updating Bayesian posterior."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1356.48,"endTime":1360.88,"body":"Every new piece of data refines the model's belief about your true trajectory."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1361.18,"endTime":1367.48,"body":"Which brings us to Pillar 3, the absolute game changer and the hardest part to engineer, the genetic integration."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1367.82,"endTime":1379.4,"body":"They incorporated 36 different polygenic risk scores, PRSs, alongside biological sex and 10 genetic principal components to rigorously account for ancestry and population stratification."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1379.7,"endTime":1382.18,"body":"This is where we fuse the static with the dynamic."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1382.44,"endTime":1385.08,"body":"Let's clarify what a polygenic risk score is first."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1384.88,"endTime":1385.1200000000001,"body":"a moment."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1385.26,"endTime":1385.9,"body":"We aren't talking about"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1386.1,"endTime":1387.3,"body":"single gene mutations here,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1387.46,"endTime":1389.02,"body":"like the BRCA mutations"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1389.3600000000001,"endTime":1390.3600000000001,"body":"for breast cancer."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1390.58,"endTime":1391.4,"body":"No, we're talking about"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1391.8,"endTime":1392.32,"body":"complex traits."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1392.88,"endTime":1393.9,"body":"Most chronic diseases,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1394.16,"endTime":1395.18,"body":"heart disease, diabetes,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1395.42,"endTime":1395.66,"body":"schizophrenia,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1396.3,"endTime":1397.22,"body":"are not caused by"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1397.22,"endTime":1398.22,"body":"a single broken gene."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1398.6200000000001,"endTime":1399.5,"body":"They are polygenic."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1399.58,"endTime":1399.68,"body":"Yeah."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1399.98,"endTime":1401.14,"body":"A polygenic risk score"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1401.32,"endTime":1402.34,"body":"looks at millions of tiny"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1402.64,"endTime":1403.48,"body":"single letter variations"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1414.8,"endTime":1414.8600000000001,"body":"of the gene."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1404.04,"endTime":1405.2,"body":"across your entire genome."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1405.82,"endTime":1408.24,"body":"Single nucleotide polymorphisms or SNPs."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1408.88,"endTime":1411.4,"body":"Each variant might only increase your risk by a fraction of a percent,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1411.68,"endTime":1415.6200000000001,"body":"but when you aggregate millions of them, you get a powerful singular baseline score"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1415.98,"endTime":1418.74,"body":"representing your innate genetic susceptibility for a trait."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1419.0,"endTime":1424.64,"body":"So, Aladdinoli takes these static polygenic risk scores, your genetic blueprint,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1425.2,"endTime":1430.3600000000001,"body":"which is identical on the day you are born and the day you die, and uses them as the anchor."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1430.64,"endTime":1438.98,"body":"They act as the gravitational pull, driving how heavily you load into those dynamic, time-varying disease signatures over the course of your life."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1439.24,"endTime":1445.6200000000001,"body":"It is an incredibly elegant architecture. The genetics drive the signatures, and the signatures drive the specific diseases."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1445.98,"endTime":1452.8,"body":"And by structuring the architecture this way, the researchers managed to solve one of the most maddening headaches in medical statistics, which you alluded to."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1452.4,"endTime":1454.88,"body":"to earlier, the competing risk problem."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1455.18,"endTime":1457.16,"body":"If we connect this to the bigger clinical picture,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1457.93,"endTime":1461.28,"body":"the competing risk problem is the bane of longitudinal studies."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1462.05,"endTime":1466.3,"body":"In a standard model, if a patient is diagnosed with one major systemic disease,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1466.6200000000001,"endTime":1468.02,"body":"say severe heart failure,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1468.48,"endTime":1472.02,"body":"The model often breaks down when trying to figure out their risk for everything else."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1472.32,"endTime":1475.92,"body":"Because the heart failure fundamentally alters their survival probability."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1476.04,"endTime":1478.26,"body":"Yes, and their interaction with the health care system entirely."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1478.94,"endTime":1482.38,"body":"But in Aladinuli, because the model is inherently generative"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1482.7,"endTime":1485.34,"body":"and because it is calculating the weighted probabilities"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1485.66,"endTime":1490.28,"body":"for 348 different diseases simultaneously through those 21 signatures,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1491.1200000000001,"endTime":1492.98,"body":"you don't just fall out of the model when you get sick."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1493.06,"endTime":1493.28,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1493.6,"endTime":1496.56,"body":"You remain at risk for the remaining 347 conditions,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1497.08,"endTime":1503.28,"body":"your overall hazard profile just elegantly mathematically updates as you absorb that new diagnosis."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1503.54,"endTime":1506.16,"body":"It recalculates the entire landscape of your future."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1506.48,"endTime":1508.14,"body":"It is like a GPS for your biology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1508.6,"endTime":1511.42,"body":"If you take a wrong turn or in this case develop a new condition,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1512.13,"endTime":1513.06,"body":"it doesn't just crash."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1513.24,"endTime":1517.52,"body":"It says recalculating and maps out the new most probable path of your biological decline."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1517.84,"endTime":1519.0,"body":"That is exactly what it does."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1519.16,"endTime":1520.48,"body":"So the Bayesian math is elegant."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1520.85,"endTime":1522.42,"body":"The theoretical architecture is sound."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1523.14,"endTime":1528.08,"body":"But Dr. Uba and her team didn't just build this in a theoretical sandbox and publish a paper about the math."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1528.41,"endTime":1531.66,"body":"They put El Adanuli to the ultimate grueling test."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1532.15,"endTime":1538.88,"body":"They deployed it across three massive, completely independent biobanks to see if it actually worked on real human lives."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1539.22,"endTime":1541.94,"body":"The scale of the validation phase here is staggering."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1542.3700000000001,"endTime":1544.2,"body":"And it's what makes this paper so authoritative."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1544.8990000000001,"endTime":1547.879,"body":"They didn't just use one local hospital data set."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1548.1490000000001,"endTime":1554.32,"body":"They utilized the UK Biobank, the Mass General Brigham Clinical Cohort, and the All of Us Research Program."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1554.659,"endTime":1558.98,"body":"We are talking about over 683,000 distinct patients."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1559.6390000000001,"endTime":1565.039,"body":"With up to 52 years of continuous longitudinal medical follow-up data,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1565.92,"endTime":1571.159,"body":"They were tracking the intertwined probabilities of 348 different diseases over half a century."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1571.46,"endTime":1573.22,"body":"This is where we get into the head-to-head battle."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1573.73,"endTime":1579.46,"body":"They pitted their magical Bayesian engine against the reigning, unquestioned champions of clinical risk scores."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1580.38,"endTime":1581.1,"body":"And they didn't cheat."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1581.659,"endTime":1583.519,"body":"This is a vital point about methodology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1584.019,"endTime":1587.56,"body":"They evaluated the model dynamically, simulating strict clinical reality."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1587.94,"endTime":1591.9,"body":"In data science, it is very easy to accidentally allow information leakage,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1592.35,"endTime":1595.5,"body":"where the model sneaks a peek at the future data to make its prediction look better."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1595.6200000000001,"endTime":1596.66,"body":"But they block that completely."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1596.92,"endTime":1598.64,"body":"Entirely. They set up strict temporal boundaries."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1599.02,"endTime":1604.76,"body":"They asked Aladinoli, based only on the genetic data and the EHR data available for this specific patient up to year 5,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1605.14,"endTime":1606.9,"body":"predict their disease trajectory in year 6."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1607.23,"endTime":1608.66,"body":"The model generates its probabilities."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1609.4,"endTime":1615.1200000000001,"body":"Then and only then, they roll the clock forward to year 6, check reality, and grade the model."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1615.88,"endTime":1619.54,"body":"Then they ask it to predict year 7 based on data up to year 6."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1620.28,"endTime":1622.0,"body":"It's a rolling dynamic evaluation."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1622.42,"endTime":1625.3,"body":"And in that strict, unforgiving, predictive environment,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1625.82,"endTime":1627.92,"body":"Elad-Nilouli absolutely dominated."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1627.96,"endTime":1631.38,"body":"It outperformed the pooled cohort equation and the newer PREVENT model"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1631.8,"endTime":1633.5,"body":"for 10-year cardiovascular predictions."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1633.8,"endTime":1637.18,"body":"It also outperformed the GL model for one-year breast cancer horizons."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1637.32,"endTime":1641.04,"body":"Across the board, they reported it achieved a median dynamic AUC area"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1641.22,"endTime":1642.6,"body":"under the curve of 0.85."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1642.999,"endTime":1649.96,"body":"For listeners who might not breathe statistics daily, the AUC is a metric that measures a classifier's ability to distinguish between classes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1650.72,"endTime":1653.84,"body":"An AUC of 0.5 means the model is basically just flipping a coin."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1654.259,"endTime":1656.259,"body":"An AUC of 1.0 is absolute perfection."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1656.6390000000001,"endTime":1656.8,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1657.08,"endTime":1662.08,"body":"In the incredibly noisy, chaotic world of predicting complex, multisystemic human health over decades..."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1662.38,"endTime":1667.64,"body":"Achieving a median dynamic AUC of 0.85 across 348 diseases is a monumental achievement."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1668.08,"endTime":1669.42,"body":"It means the model is highly robust."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1669.8990000000001,"endTime":1673.76,"body":"But we have to address a massive elephant in the room regarding the data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1674.82,"endTime":1676.96,"body":"No real-world data set is perfect."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1677.7,"endTime":1682.3,"body":"And the team had to pull off a deeply impressive, rigorous mathematical triumph"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1683.019,"endTime":1687.72,"body":"to make sure their stunning results weren't just a mirage caused by biased data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1687.78,"endTime":1701.98,"body":"This raises a critical issue about the fidelity of biobank data, particularly the UK biobank. It is a phenomenal resource, one of the best in the world, but it suffers from a well-documented epidemiological phenomena known as participation bias."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1702.4,"endTime":1703.72,"body":"The healthy volunteer effect."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1703.94,"endTime":1704.44,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1704.88,"endTime":1709.96,"body":"The people who have the time, resources, and inclination to volunteer for long-term intensive"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1710.1200000000001,"endTime":1715.3,"body":"health tracking studies tend to be, on average, significantly healthier, wealthier, and more"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1715.42,"endTime":1717.08,"body":"health conscious than the general population."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1717.28,"endTime":1722.3600000000001,"body":"So if you train your AI exclusively on a population of super healthy, health conscious volunteers,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1723.08,"endTime":1726.24,"body":"the model is going to learn artificially low baseline hazards."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1726.52,"endTime":1730.38,"body":"It's going to be fundamentally miscalibrated when you try to deploy it in a real world general"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1730.58,"endTime":1731.06,"body":"public clinic."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1731.3600000000001,"endTime":1731.96,"body":"Precisely."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1732.2,"endTime":1734.7,"body":"the baseline probabilities would be skewed."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1735.159,"endTime":1739.22,"body":"This is a violation of the missing-at-random assumption in statistics."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1739.8,"endTime":1741.58,"body":"If they just ran the data raw,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1742.18,"endTime":1745.119,"body":"Aladinoli would learn a distorted version of human biology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1745.379,"endTime":1746.3,"body":"Which defeats the purpose."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1746.619,"endTime":1750.379,"body":"Right. But because Aladinoli is a generative model"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1750.3990000000001,"endTime":1753.98,"body":"with an explicit mathematically defined likelihood formulation,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1754.84,"endTime":1761.26,"body":"They were able to implement a rigorous statistical correction called inverse probability weighting, or IPW."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1761.54,"endTime":1764.26,"body":"Let's break down exactly how IPW rescues the math here."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1764.72,"endTime":1768.66,"body":"IPW modifies the actual likelihood function during the model's training phase."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1769.17,"endTime":1772.58,"body":"It assigns an observation weight to every single patient in the training set."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1773.26,"endTime":1782.96,"body":"If a patient belongs to a demographic that we know is overrepresented in the biobank, say, wealthy, healthy individuals, IPW mathematically downweights their contribution to the final probability curve."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1783.58,"endTime":1788.08,"body":"Conversely, if a patient belongs to a demographic that is underrepresented, their data is upweighted."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1788.26,"endTime":1791.78,"body":"So by mathematically adjusting the weights and the likelihood function..."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1791.78,"endTime":1800.28,"body":"they force the model to learn a probability distribution that accurately reflects the true underlying biological signals of the broader population,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1800.98,"endTime":1803.42,"body":"rather than the skewed demographics of the biobank."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1803.6,"endTime":1806.64,"body":"Exactly. It prevents the model from suffering asymptotic bias."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1807.46,"endTime":1819.34,"body":"It is a rigorous, necessary step to ensure that when Al Adinley says you have a 40 percent risk, that number is actually calibrated to reality, not to a fantasy world of perfectly healthy volunteers."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1819.779,"endTime":1825.3,"body":"Okay, so they built a mathematically rigorous, unbiased model that predicts the future better than the clinical gold standard."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1825.28,"endTime":1827.68,"body":"standards. That alone is worthy of a deep dive."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1828.44,"endTime":1831.28,"body":"But here is the massive \"aha\" moment for me in this paper."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1831.96,"endTime":1836.24,"body":"Elginouli isn't just a crystal ball for doctors, it is a microscope for biologists."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1836.5,"endTime":1839.3600000000001,"body":"This is where the dual nature of the Bayesian framework truly shines."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1839.84,"endTime":1844.5,"body":"It predicts the future, yes, but because it explicitly models those 21 latent signatures,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1844.9,"endTime":1846.88,"body":"it also maps the underlying biology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1847.33,"endTime":1848.72,"body":"It becomes an engine for discovery."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1849.14,"endTime":1851.8600000000001,"body":"Because it maps these underlying biological signatures,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1852.43,"endTime":1855.02,"body":"the team realized they could run a new kind of genetic study."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1855.82,"endTime":1858.56,"body":"Instead of just looking at how genetics map to a single disease,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1858.89,"endTime":1862.8,"body":"which is how standard genome-wide association studies, or GLEDs, work,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1863.44,"endTime":1865.58,"body":"they looked at how genetics mapped to the signatures."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1866.02,"endTime":1871.08,"body":"They ran a signature-based G-violence, and the results were incredibly revealing."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1871.3600000000001,"endTime":1875.94,"body":"They discovered 151 genome-wide significant LOSAR."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1876.42,"endTime":1882.44,"body":"These are 151 specific locations on the human genome that are deeply associated with disease risk."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1882.8,"endTime":1884.539,"body":"But here's the crazy part."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1885.159,"endTime":1888.6390000000001,"body":"Traditional single trait analyses completely missed some of these leci."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1888.92,"endTime":1890.759,"body":"Scientists had been staring right at them,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1890.759,"endTime":1894.96,"body":"but because they were only looking for a direct line between a gene and one specific disease,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1895.289,"endTime":1898.22,"body":"like DNA versus heart attack, they couldn't see the signal."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1898.82,"endTime":1903.019,"body":"But because Aladinoli zooms out and looks at the entire signature of the neighborhood of diseases,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1903.74,"endTime":1906.039,"body":"it spotted the genetic culprits hiding in plain sight."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1906.42,"endTime":1908.94,"body":"It proves a concept in genetics called pleiotropy."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1909.3,"endTime":1913.22,"body":"Pleiotropy meaning that a single genetic variant exerts effects on multiple different traits"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1913.25,"endTime":1914.5,"body":"or diseases simultaneously."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1915.289,"endTime":1915.42,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1916.47,"endTime":1922.279,"body":"If a specific genetic variant subtly increases your inflammation, it might slightly increase"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1922.289,"endTime":1927.019,"body":"your risk for arthritis, and slightly increase your risk for heart disease, and slightly increase"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1927.07,"endTime":1928.1200000000001,"body":"your risk for psoriasis."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1928.1200000000001,"endTime":1928.5,"body":"Oh, I see."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1928.74,"endTime":1933.519,"body":"If you look at any one of those diseases in isolation, the genetic signal is too weak to"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1933.58,"endTime":1935.7,"body":"cross the threshold of statistical significance."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1936.19,"endTime":1936.8600000000001,"body":"It looks up noise."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1937.52,"endTime":1943.68,"body":"But aladynulaya aggregates those subtle signals across the entire inflammatory signature."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1944.08,"endTime":1946.84,"body":"And suddenly the signal becomes massive and undeniable."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1947.32,"endTime":1949.22,"body":"Alladynly doesn't just catch the final disease."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1949.42,"endTime":1950.76,"body":"It catches the mechanism."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1951.08,"endTime":1958.56,"body":"And that leads to this incredible paradigm shattering concept of biological subtypes or heterogeneity within disease categories."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1959.039,"endTime":1965.72,"body":"What is truly fascinating here is that the model proves mathematically that our current diagnostic labels are often woefully inadequate."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1965.92,"endTime":1966.22,"body":"How so?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1966.68,"endTime":1969.32,"body":"Let's ground this in a hypothetical scenario that happens every day."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1969.44,"endTime":1971.46,"body":"You have two different patients, patient A and patient B."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1972.26,"endTime":1976.5,"body":"Both arrive at the emergency room and both suffer a myocardial infarction, a heart attack."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1977.039,"endTime":1978.94,"body":"On their medical chart, the diagnosis is identical."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1979.0,"endTime":1980.5,"body":"The ICD-10 code is identical."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1980.8990000000001,"endTime":1988.3600000000001,"body":"But Alati Newley reveals that their biological journey to that exact same hospital bed could be entirely fundamentally different."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1988.2,"endTime":1992.98,"body":"different. Entirely different. Because Aladinili maps their signature loadings over time,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1993.58,"endTime":1998.6390000000001,"body":"we can see the historical engine of their disease. The model shows that patient A might have arrived"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":1998.69,"endTime":2004.1390000000001,"body":"at that heart attack through a heavily metabolic pathway. Right. For 20 years, their risk was driven"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2004.34,"endTime":2010.019,"body":"by escalating insulin resistance, obesity, and diabetic complications that slowly destroyed"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2010.18,"endTime":2017.58,"body":"their vascular system. While patient B might have arrived at the exact same heart attack through a"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2017.6,"endTime":2022.44,"body":"pathway, maybe they have zero metabolic issues, perfect blood sugar, but they have a massive"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2022.88,"endTime":2027.039,"body":"chronic inflammatory loading driven by a specific polygenic risk score."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2027.44,"endTime":2030.22,"body":"And the model can separate these patients with astonishing clarity."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2030.68,"endTime":2034.5,"body":"The paper noted that the statistical effect sizes when comparing these patient clusters"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2034.6,"endTime":2035.159,"body":"were massive."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2035.92,"endTime":2038.3,"body":"They cited a Cohen's D of up to 4.25."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2038.82,"endTime":2042.82,"body":"To put that in perspective for the listener in statistics, a Cohen's D measures the standardized"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2043.08,"endTime":2044.06,"body":"difference between two means."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2044.6200000000001,"endTime":2047.8600000000001,"body":"A score of a 0.8 is generally considered a large effect size."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2048.18,"endTime":2049.0,"body":"It's very large."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2049.2889999999998,"endTime":2059.98,"body":"A 4.25 means the biological signatures of these patients, despite having the exact same clinical diagnosis, are separated by vast, undeniable canyons of biological difference."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2060.5899999999997,"endTime":2067.519,"body":"The researchers noted that over 95% of their comparisons between these subtypes yielded highly significant P-virus."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2067.1,"endTime":2070.94,"body":"values is a profound realization yeah you have two people with the same disease"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2071.36,"endTime":2075.42,"body":"yeah but they likely require completely different preventative strategies and"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2075.58,"endTime":2079.32,"body":"treatments because the biological engine driving their disease is different wow"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2079.68,"endTime":2083.96,"body":"if you give patient B a drug designed to fix patient A's metabolic pathway it"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2083.96,"endTime":2088.379,"body":"might do absolutely nothing Aleta Neely maps the engine it shifts the entire"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2088.78,"endTime":2092.98,"body":"paradigm from treating a generalized label to treating a specific patient level"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2093.18,"endTime":2093.78,"body":"biological process"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2094.2,"endTime":2094.98,"body":"Yes, it really does."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2095.36,"endTime":2102.14,"body":"Now, to show that this isn't just one brilliant team in Boston having a eureka moment with Bayesian math, we have to talk about the tech giants."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2102.61,"endTime":2115.68,"body":"Because while Dr. Erbet's team was perfecting Eleni Newley, Silicon Valley was looking at the exact same massive problem merging messy EHRs with static genetics and deciding to throw overwhelming brute force computing power at it."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2115.84,"endTime":2123.22,"body":"Right. If we shift our focus to the paper titled Integrating Genomics into Multimodal EHR Foundation Model,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2123.98,"endTime":2128.06,"body":"which was authored by a massive collaboration between Verily, Google, and NVIDIA."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2128.56,"endTime":2130.66,"body":"They tackled the exact same biological problem,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2131.1,"endTime":2133.72,"body":"but they took a fundamentally different architectural approach."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2134.14,"endTime":2135.68,"body":"They didn't use Bayesian mixtures."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2135.92,"endTime":2138.72,"body":"They used large language models, LLMs."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2139.04,"endTime":2144.94,"body":"They took the exact same underlying transformer architecture that powers tools like ChatGPT,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2145.41,"endTime":2150.94,"body":"but instead of predicting the next word in a sentence, they are trying to predict the next disease in your medical chart."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2151.22,"endTime":2153.82,"body":"They treat the medical record literally as a language."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2154.5,"endTime":2157.22,"body":"A diagnosis is a word. A hospital visit is a sentence."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2157.78,"endTime":2160.18,"body":"Your entire medical history is a document."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2160.68,"endTime":2162.079,"body":"It's a crazy way to think about it."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2162.2,"endTime":2163.16,"body":"It is, but it works."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2163.86,"endTime":2167.92,"body":"They pre-train a massive foundation model on millions of these clinical trajectories,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2168.559,"endTime":2172.18,"body":"essentially teaching the neural network the complex grammar and syntax of human health."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2172.42,"endTime":2174.099,"body":"But the challenge they faced was immense."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2174.88,"endTime":2179.62,"body":"If you are building an LLM that reads a timeline, a sequence of events,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2180.6,"endTime":2183.86,"body":"How do you inject static, unchanging DNA into it?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2184.56,"endTime":2195.7400000000002,"body":"If I hand an LLM a book to read chapter by chapter, how do I simultaneously force it to inject a permanent overarching theme or vibe into its understanding of every single page?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2195.92,"endTime":2197.82,"body":"It is a profound engineering challenge."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2198.36,"endTime":2204.279,"body":"You have a dynamic time series modality on one hand and a static high-dimensional genomic modality on the other."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2204.9,"endTime":2205.62,"body":"How do you fuse them?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2206.42,"endTime":2209.819,"body":"The Verily team solved this using sophisticated neural network techniques,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2210.599,"endTime":2213.059,"body":"specifically adapter modules and cross-attention mechanisms."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2213.42,"endTime":2216.46,"body":"Let's focus on the cross-attention because the mechanics of how this works are wild."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2216.72,"endTime":2219.96,"body":"How does the model actually see the DNA while reading the hospital record?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2220.2,"endTime":2224.279,"body":"Cross-attention is a mechanism borrowed from encoder-decoder paradigms in machine translation."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2224.84,"endTime":2231.06,"body":"First, they take the polygenic risk scores, the static genetic data, and they project them through a dedicated neural network layer."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2231.67,"endTime":2233.94,"body":"This projection creates what they call soft tokens."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2234.32,"endTime":2237.1,"body":"Soft tokens. What does that mean in the context of an LLM?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2237.34,"endTime":2244.36,"body":"Think of the model's vocabulary. It has hard tokens for actual medical concepts, like an ICD code for asthma."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2245.2400000000002,"endTime":2250.299,"body":"A soft token is essentially a synthetic continuous vector in the model's embedding space."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2250.5190000000002,"endTime":2251.16,"body":"Okay, I follow."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2251.2,"endTime":2253.86,"body":"It doesn't correspond to a specific word in the medical dictionary."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2254.539,"endTime":2258.339,"body":"Instead, it is a dense mathematical essence of your genetic risk."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2258.559,"endTime":2262.7,"body":"So they turn your DNA into a mathematical vibe that the language model can understand."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2263.059,"endTime":2263.259,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2263.539,"endTime":2268.0,"body":"Now, as the transformer model reads through your clinical history chronologically, step"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2268.0,"endTime":2270.7,"body":"by step, the cross-attention mechanism kicks in."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2271.14,"endTime":2273.64,"body":"It uses a query key value system."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2273.8,"endTime":2274.66,"body":"Like a database search."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2275.0,"endTime":2275.38,"body":"Basically."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2276.09,"endTime":2279.46,"body":"The patient's current clinical state, where they are in their medical history right now,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2279.66,"endTime":2280.34,"body":"acts as the query."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2281.06,"endTime":2283.34,"body":"The genetic soft tokens act as the keys and values."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2283.68,"endTime":2288.06,"body":"So at every single step of your timeline, as the model reads the story of your life,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2288.45,"endTime":2292.82,"body":"the cross-attention mechanism mathematically forces the model to constantly attend to"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2293.14,"endTime":2297.56,"body":"or reference those genetic soft tokens to see how the story should unfold next."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2297.9,"endTime":2299.46,"body":"It's a continuous dynamic fusion."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2299.96,"endTime":2304.299,"body":"The model asks, \"Given that the patient just developed hypertension in the query,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2305.12,"endTime":2308.819,"body":"and given their specific underlying genetic risk profile for cardiovascular disease,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2309.559,"endTime":2313.0190000000002,"body":"the key value what is the most mathematically probable next event?\""},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2313.059,"endTime":2316.68,"body":"And did this wildly complex cross-attention mechanism actually work?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2317.039,"endTime":2321.559,"body":"Phenomenally well. They evaluated this multimodal architecture on the All of Us dataset"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2322.259,"endTime":2326.299,"body":"when they explicitly injected that genetic data into the foundation model via cross-attention,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2326.96,"endTime":2333.14,"body":"They saw a massive 15.5% relative increase in the AUPRC, the area, under the precision"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2333.4,"endTime":2337.66,"body":"recall curve, specifically for predicting the onset of type 2 diabetes within a 10-year"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2337.8,"endTime":2337.92,"body":"window."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2338.319,"endTime":2342.9,"body":"A 15.5% jump in predictive precision just by giving the LLM the genetic content."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2342.92,"endTime":2344.0,"body":"It is a massive leap."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2344.089,"endTime":2348.559,"body":"It proves definitively that relying on EHR data alone, even with a massive foundation"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2348.91,"endTime":2351.12,"body":"model, leaves critical predictive power on the table."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2351.559,"endTime":2355.259,"body":"Combining longitudinal clinical data with static genomic data is unequivocally the holy"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2355.46,"endTime":2356.299,"body":"grail of predictive modeling."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2356.559,"endTime":2359.4,"body":"Here's where it gets really interesting for me regarding the Verily paper."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2360.0190000000002,"endTime":2361.72,"body":"It wasn't just about combining the data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2362.349,"endTime":2366.819,"body":"They also invented a breakthrough in how they actually calculate the final risk percentage."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2367.52,"endTime":2369.44,"body":"They call it path computing probabilities."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2369.88,"endTime":2375.44,"body":"Yes, this is a crucial evaluation from how early, naive EHR foundation models function."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2376.22,"endTime":2381.12,"body":"Previously, if you wanted to predict a future risk, you would use something called Monte Carlo sampling."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2381.28,"endTime":2382.7400000000002,"body":"Which means rolling the dice."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2382.86,"endTime":2383.84,"body":"over and over again."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2384.02,"endTime":2384.1,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2384.3,"endTime":2385.92,"body":"You have the patient's history up to today."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2386.18,"endTime":2388.06,"body":"You ask the LLM to generate, say,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2388.54,"endTime":2390.94,"body":"100 possible synthetic futures for that patient"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2391.2400000000002,"endTime":2392.54,"body":"based on the learned probabilities."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2392.7400000000002,"endTime":2392.96,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2393.2,"endTime":2394.88,"body":"If the patient develops diabetes in 20"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2395.04,"endTime":2396.82,"body":"out of those 100 simulated futures,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2397.42,"endTime":2399.58,"body":"the model assigns a 20% risk score."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2399.76,"endTime":2402.36,"body":"But that creates really clunky, noisy,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2402.8,"endTime":2403.88,"body":"bucketed risk scores."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2404.539,"endTime":2409.44,"body":"Because it's discrete sampling, you get 20% or 21%, but it's not a smooth mathematical truth."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2409.549,"endTime":2412.339,"body":"It's an approximation based on how many times you rolled the dice."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2413.0190000000002,"endTime":2418.22,"body":"Exactly. It is computationally expensive and difficult to use for fine-grained clinical risk stratification."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2419.099,"endTime":2422.68,"body":"Path computing probabilities completely bypass the Monte Carlo dice rules."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2423.42,"endTime":2431.779,"body":"Instead of generating distinct futures, the algorithm uses exact Bayesian update rules at every single token generation step along the path."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2431.98,"endTime":2439.92,"body":"So it mathematically calculates the exact probability of every specific incremental step happening, multiplying the sequential probabilities together."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2440.089,"endTime":2447.94,"body":"Yes. It calculates the cumulative exact probability of a target token, like a diabetes diagnosis, occurring at any point in the future sequence."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2448.64,"endTime":2455.28,"body":"This generates a highly nuanced, completely continuous, exact mathematical curve of unique risk for every single individual."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2455.68,"endTime":2459.52,"body":"It extracts so much more meaningful, high-resolution signal from the data."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2459.98,"endTime":2468.42,"body":"It allows a hospital system to look at a patient and set incredibly precise, individualized thresholds for intervention, rather than relying on clunky 10% buckets."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2468.52,"endTime":2472.6,"body":"It represents a massive leap in the mathematical maturity of foundation models in health care."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2472.86,"endTime":2486.42,"body":"So we have to pull all these threads together. What does this all actually mean? We have Dr. Erbit's team building a Latinoi, identifying hidden biological signatures and disease engines with elegant Bayesian math."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2486.8,"endTime":2487.16,"body":"Yes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2487.21,"endTime":2496.48,"body":"And we have Google and Verily using massive large language models with cross attention to predict the next clinical word in your chart with astonishing accuracy."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2496.84,"endTime":2502.26,"body":"What does this convergence mean for you, the listener, sitting in a doctor's office five or ten years from now?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2502.5,"endTime":2511.14,"body":"If we connect all of this to the bigger picture, it signals the definitive end of reactive acute intervention as the primary mode of health care."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2511.84,"endTime":2515.22,"body":"We are moving irreversibly toward continuous proactive health management."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2515.67,"endTime":2521.3,"body":"The synthesis of these technologies effectively allows the health care system to create a predictive digital twin of you."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2521.56,"endTime":2524.76,"body":"A digital twin. That sounds like sci-fi, but it's mathematically real."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2524.8,"endTime":2537.0,"body":"It is. A virtual, high-dimensional representation of your biology that is constantly, dynamically updated with every single lab test, every new prescription, every minor doctor's visit, all permanently grounded by your unique genetic code."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2537.3,"endTime":2541.22,"body":"Your physician won't just be looking at the physical use sitting on the exam table."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2541.76,"endTime":2548.62,"body":"They will be consulting your digital twin to see exactly where your current biological trajectory is mathematically leading."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2548.95,"endTime":2554.28,"body":"And think about how this completely overhauls the pharmaceutical industry and how we run clinical trials."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2555.52,"endTime":2558.38,"body":"Right now, drug development is incredibly inefficient."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2558.96,"endTime":2567.98,"body":"If I have a promising new experimental drug for breast cancer, I design a trial and look for a bunch of patients who have the broad label breast cancer on their chart."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2568.599,"endTime":2578.0,"body":"But as we discussed deeply with the myocardial infarction subtypes, that broad clinical label masks massive underlying biological heterogeneity."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2578.2,"endTime":2578.3,"body":"Right."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2578.64,"endTime":2586.26,"body":"If you test a drug on a heterogeneous group, the signal of efficacy often gets drowned out by the noise of patients who have a different underlying disease engine."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2586.84,"endTime":2587.7,"body":"The trial fails."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2587.98,"endTime":2592.96,"body":"But with L. DeNoli's continuous signature loadings, we don't have to rely on broad labels anymore."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2593.14,"endTime":2595.26,"body":"We can do signature-based patient matching."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2595.66,"endTime":2601.78,"body":"I can match you with 500 other patients who share your exact, specific, underlying biological engine."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2602.62,"endTime":2609.7,"body":"Say, a highly specific intersection of the metabolic and inflammatory signatures driven by a specific polygenic risk score."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2610.16,"endTime":2614.54,"body":"I can build a clinical trial based on the pathway, not the final endpoint."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2614.92,"endTime":2620.06,"body":"It could dramatically increase the success rate of clinical trials because the pharmaceutical"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2620.44,"endTime":2627.0,"body":"interventions would be hyper-targeted at the actual verified mechanism of disease for that"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2627.2400000000002,"endTime":2629.04,"body":"specific mathematically defined subgroup."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2629.5,"endTime":2633.34,"body":"We would waste less time, spend less money, and bring effective targeted cures to patients"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2633.98,"endTime":2634.88,"body":"significantly faster."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2634.88,"endTime":2638.48,"body":"But I have to ask one final critical question regarding the data foundation of all"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2638.5,"endTime":2643.18,"body":"this. Both the Bayesian models and the LOMs rely on massive amounts of data to learn these"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2643.46,"endTime":2648.38,"body":"probabilities. Extensive medical histories, detailed genetic sequencing, decades of follow-up."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2648.86,"endTime":2651.7400000000002,"body":"How do we ensure that these models are actually accurate for everyone?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2652.06,"endTime":2652.94,"body":"That's a vital question."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2653.2,"endTime":2656.92,"body":"If a model is trained primarily on data from populations who have excellent healthcare"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2657.18,"endTime":2660.9,"body":"access and frequent doctor visits, aren't we just building a mathematical reality that"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2660.98,"endTime":2661.64,"body":"only works for them?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2661.94,"endTime":2665.64,"body":"That is the most critical hurdle for the clinical deployment of these models."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2666.0,"endTime":2668.78,"body":"It is a matter of pure data fidelity and calibration."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2669.81,"endTime":2681.22,"body":"If a model, particularly one relying on polygenic risk scores, is trained exclusively on genomic data from individuals of European descent, it will suffer from severe population stratification errors."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2681.52,"endTime":2687.88,"body":"The model will simply be mathematically blind to the genetic variations that drive disease in other populations."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2688.36,"endTime":2693.2,"body":"Precisely. The baseline probabilities will be fundamentally miscalibrated for individuals"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2693.39,"endTime":2698.2400000000002,"body":"of non-European ancestries, rendering the predictions inaccurate and clinically useless"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2698.25,"endTime":2703.96,"body":"for those patients. This is not just a theoretical concern. It is a known flaw in early genomic"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2704.069,"endTime":2704.259,"body":"research."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2704.64,"endTime":2707.48,"body":"So how do the researchers solve this calibration problem?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2707.5,"endTime":2709.7,"body":"You solve it by fixing the training data itself."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2710.08,"endTime":2715.42,"body":"This is why both of our primary sources explicitly highlight their reliance on datasets like the"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2715.46,"endTime":2718.0,"body":"All of Us Research Program and the Errigen Network."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2718.28,"endTime":2722.14,"body":"The All of Us cohort was central to validating the Verily Foundation model, right?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2722.42,"endTime":2724.58,"body":"Yes, and the Al Genoey team utilized it as well."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2725.3,"endTime":2730.94,"body":"The All of Us program is a massive nationwide initiative specifically engineered to capture"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2731.3,"endTime":2734.34,"body":"comprehensive data from historically underrepresented populations."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2734.7400000000002,"endTime":2739.319,"body":"They intentionally bypass the standard homogenous biobank demographics."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2740.02,"endTime":2744.2,"body":"They recruit broadly to build a truly diverse multi-ancestry cohort."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2744.559,"endTime":2747.66,"body":"And the Emerge consortium serves a similar structural purpose."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2748.059,"endTime":2748.42,"body":"Exactly."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2749.16,"endTime":2755.2,"body":"Emerge focuses heavily on the real-world clinical implementation of genomic data across diverse patient populations."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2756.14,"endTime":2771.2400000000002,"body":"By actively choosing to train and calibrate L.D. Nolley and these massive foundation models on diverse, highly representative data sets, the researchers can mathematically ensure that the predictive probabilities are robust, accurate, and scientifically valid across a wider spectrum of human biology."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2771.68,"endTime":2775.98,"body":"Okay, let's take a deep breath and recap the incredible intellectual journey we have been on today."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2776.0,"endTime":2780.76,"body":"today. We started in a frustrating world of messy, isolated snapshots. Doctors looking"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2780.76,"endTime":2784.58,"body":"at a single blood test and trying to guess the future using static equations like the"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2784.68,"endTime":2787.72,"body":"PCE that mathematically failed the moment a patient's life changes."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2788.3,"endTime":2794.08,"body":"And we explore the immense, unglamorous difficulty of cleaning that clinical data, relying on"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2794.18,"endTime":2800.28,"body":"sophisticated vector spaces and pipelines like PERT to harmonize the linguistic chaos"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2800.6,"endTime":2803.06,"body":"into a structure that machines can actually comprehend."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2803.46,"endTime":2812.1,"body":"Then we met the team behind Alade Nulli, blending magic and math, using a rigorous Bayesian framework to treat your health like a dynamic buffet of probabilities,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2813.12,"endTime":2821.38,"body":"uncovering hidden biological signatures, proving the existence of distinct disease subtypes, and mathematically outperforming the clinical gold standard."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2821.78,"endTime":2838.32,"body":"And we saw how the tech giants are pushing the envelope from a different angle, using large language models and intricate cross-attention mechanisms to seamlessly inject static DNA into the dynamic, continuous timeline of your life, proving definitively that multimodal integration is the future of precision medicine."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2838.64,"endTime":2849.94,"body":"All of this research is driving toward a reality where you possess a predictive digital twin, and medicine evolves to treat the underlying engine of your disease long before the ultimate clinical label is ever applied."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2850.16,"endTime":2854.96,"body":"It is a profound, irreversible structural transition in our fundamental understanding"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2855.02,"endTime":2855.6,"body":"of human health."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2855.8,"endTime":2859.8,"body":"Which leaves me with a final lingering thought for you to ponder as we wrap up this deep dive."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2860.68,"endTime":2866.34,"body":"If we are entering an era where a machine can seamlessly integrate your static DNA and"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2866.34,"endTime":2870.92,"body":"the semantic context of every single doctor's visit you've ever had, in order to predict"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2870.96,"endTime":2877.02,"body":"the exact continuous trajectory of your biological decline with unprecedented mathematical accuracy,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2878.14,"endTime":2881.22,"body":"Does knowing that trajectory rob you of the illusion of control?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2882.32,"endTime":2885.9,"body":"Or does seeing the jagged white line years before the bone even breaks"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2886.3,"endTime":2890.54,"body":"give you the ultimate cheat code to actively, aggressively rewrite your own fate?"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2890.72,"endTime":2893.84,"body":"That is the ultimate question at the frontier of predictive medicine."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2894.28,"endTime":2896.0,"body":"Thank you so much for joining us on this deep dive."},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2896.26,"endTime":2897.78,"body":"The Future of Medicine is being written today,"},{"speaker":"Right. I don't plug myself into a diagnostic bay every Tuesday at 9","startTime":2898.16,"endTime":2901.04,"body":"and we loved unpacking the incredible mechanisms behind it with you."}]}