Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
David Shaywitz
The Great Technology Plan for Wellness, to simplify only slightly, envisions a path to health that runs through data: collect more (through wearables and blood tests), analyze with AI, and deliver personalized coaching that improves as the data grow and the models sharpen.
It sounds sensible, and one day it may even deliver — I’m long-term optimistic.
But today it seems largely precision health theater; our ability to offer advice that is both meaningfully personalized and scientifically valid is remarkably limited, as Kevin Hall persuasively demonstrated in the case of continuous glucose monitors for non-diabetics (the instruments are unquestionably valuable for those with diabetes).
In fact, if you use tech wellness platforms as directed — intensively pursuing metric maximization — there’s a decent chance you’ll wind up feeling worse. A slew of adverse consequences are starting to be described, from orthosomnia (trouble sleeping because you’re obsessed with sleep metrics) to orthorexia (the obsessive pursuit of “perfect” eating). A New York Times feature introduced us to “Oura paranoia,” a debilitating obsession with the data from your wearables; one user described it as a constant effort to hit metrics in order to “make the ring happy.”
Or consider how Rachel Feinzeig described her time on her exercise bike: “I get on my Peloton and am confronted with a veritable dashboard of my inadequacies: cadence (sluggish), resistance level (embarrassing), output (am I even alive?).”
Do these platforms actually enhance health? So far, the data aren’t compelling. Several comprehensive academic reviews — one in 2022, another in 2025 — found solid evidence of wearables increasing physical activity (slightly), but mixed results on actual health impact.
If there is a benefit, it may come not from data-driven optimization but from the effect of these platforms on mindset — the degree to which they reinforce your sense that you’re the sort of person who cares about their health, and supply dynamic visible evidence of your commitment to it.
Mindset, as a mechanism, is surprisingly powerful. A classic 2007 Harvard study examined hotel workers who viewed themselves as doing very little activity. One group was told (accurately) that their daily work already met the criteria for an active lifestyle; the other was told nothing. Four weeks later, the informed group had significantly reduced their body weight, body fat, and blood pressure compared to the control group — apparently driven simply by the shift in mindset, since their actual activity did not change.
The impact of self-belief shows up at the population level too, across a much longer horizon. Examining national samples of over 60,000 adults followed for up to 21 years, researchers found that people who believed they were less active than others their age were far more likely to be dead at follow-up — even after adjusting for a range of risk factors, including actual activity. Believing yourself sedentary independently predicted earlier death, even after accounting for how much you actually moved. To be clear, moving more genuinely is associated with better health; the surprise is that mindset adds to it, independent of the activity itself.
A large CVS/Aetna analysis points in the same direction. Members enrolled in a wearable-and-incentive program incurred in the first year roughly $10 less in monthly medical spending than matched non-enrollees — and the savings came almost entirely from fewer non-emergent emergency room visits and reduced specialist and mental-health visits. That looks less like bodies that grew healthier than like people who found themselves in a better headspace and made different decisions about when and where to seek care.
But belief cuts both ways. In a 2023 study, people wearing an Apple Watch received manipulated feedback about their walking: one group saw a step count deflated by 40%, another saw it inflated by 40%, and a control group saw the truth — about 7,000 steps for all of them, well above the US background level of around 4,000. Compared to controls, the deflated group (seeing ~4,200) ate worse, felt worse, and had higher blood pressure and heart rates after four weeks. The inflated group (~9,800) didn’t differ from controls — probably because 7,000 and 9,800 read as much the same, while 4,200 reads as meaningfully low.
What stands out is what didn’t move. The belief reliably changed how people felt, what they ate, and how their bodies registered — heart rate, blood pressure — while the walking itself stayed essentially flat. The mindset shifted; the behavior it was ostensibly about did not follow.
The sports performance literature, by contrast, is replete with cases where inflated feedback shifts mindset and behavior together. Endurance runners given fake EPO ran measurably faster; runners told (falsely) they’d scored above average lifted their VO₂max on retest; cyclists on placebo ergogenics produced measurably more power — gains generally in the 1–5% range. Inflated feedback seems to work when it targets a salient, actionable performance gap, which “7,000 versus 9,800 steps” apparently wasn’t.
Mindset, then, reliably moves how you feel, how you eat, and how your body responds. Whether it moves what you actually do is a separate question — sometimes it does, as the large wearable reviews and the sports studies show, and sometimes, as with those manipulated step counts, it doesn’t. A changed mind doesn’t automatically become a changed life, and belief still has to be translated into durable action.
The initial challenge is how to influence mindset — which is where Dr. Peter Attia, author of Outlive, joins the story. Attia, as best I can tell, has motivated far more people than the many sensible physician- and scientist-authors who urge readers to eat deliberately, move joyfully, and connect with others — exactly the right advice. Attia urges much the same thing, but wraps it in exceptionally specific, detailed protocols, a concrete, actionable-sounding roadmap that gives people something to grip in a way the other authors, very reasonably, do not.
Yet the readers likely to benefit most from Attia aren’t the ones who adhere to his script with religious zeal; it’s those who absorb the spirit and believe they now have a place to start — who take prevention more seriously even if they never get the comprehensive bloodwork he favors, for example. Attia provides a useful scaffold that falters when it’s taken as an inviolable prescription.
Many wellness tech platforms, I suspect, work the same way. For users engaging with these platforms in the most healthy fashion, quantification isn’t really delivering precision so much as proof. A step count takes an ordinary walk and turns it into evidence — evidence you can see, and show, that you’re the sort of person who moves and pays attention. The hotel workers got that validation in a sentence, but most of us would wave a sentence away; we trust a number on a screen far more than a pep talk – perhaps because a number aligns better with our technology-inflected explanatory model of health.
For many people, the palpability of this receipt is what makes the effort feel real enough to keep making. That is what these platforms can offer — not the optimization they advertise, but the sense of being someone who takes their health seriously, and a place to act on it. (Supplements may work in much the same way, contributing more directly to identity than biology.)
It’s also worth noting that while so much of today’s data-driven personalization looks like theater, the same companies patiently collecting reams of biometric data may be building something genuinely valuable: a longitudinal asset that assesses each individual against their own trajectory rather than against population norms. This is the approach Nathan Price, Professor and Co-Director of the Center for Human Healthspan at the Buck Institute and Chief Scientific Officer of Thorne, has championed for years, and described in his prescient 2023 book with Lee Hood, The Age of Scientific Wellness.
At some point we’ll hit the inflection between theater and real value. Until then, if we’re going to embrace tech wellness platforms, let’s make sure we’re not holding on too tightly — and that the platforms are serving us, rather than the other way around.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Woody Sherman, chief innovation officer, PsiThera; Chair of the OpenFold Consortium Executive Committee
In wet labs and dry labs, investor meetings, and medical conferences, it’s impossible to avoid discussions about how AI is transforming the way medicines are discovered.
It is and it isn’t.
This “AI in drug discovery” dogma gets something wrong. It assumes that “AI in drug discovery” is one thing. It is not. It’s true that AI systems help drug discovery teams work faster by organizing information, automating tasks, and, most tantalizingly, generating plausible ideas for future medicines at unprecedented speed.
But there’s still a lot missing. This is language-based AI. Faster, more organized work through large language models is not the same as better prediction of the properties of a physical structure in a live organism. The hardest question in drug discovery remains stubbornly tied to the physical world, not the digital one: which molecules will actually work in biology’s messy, dynamic environment?
To materially improve drug discovery, we need to distinguish between systems that organize information with language and systems that predict physical reality. Until the field makes the distinction between orchestrating and predicting, capital and scientific effort will continue to flow toward the wrong problems, and patients will wait longer than they should for new medicines.
The most visible AI systems today — including ChatGPT, Claude, and Gemini — are built to recognize patterns in symbolic information such as alphabets, sequences, and other structured data. They are already proving useful in drug discovery, from helping teams search information, summarizing knowledge to generating hypotheses and working more efficiently.
What makes drug discovery hard is that every promising molecule must pass through a gauntlet of physical tests rooted in chemistry, biology, and pharmacology. What matters is the physical properties of these molecules in the real world, not just in how language can describe them. Language- based AI can help summarize and classify what we already know. What it cannot do, at least not on its own, is reliably predict the physical properties that determine whether a molecule will actually work in a cell assay, animal model, or human patient. That is the gap Physical AI is meant to close.
An AI system trained on airplane blueprints might generate plausible aircraft designs, but engineers would never skip aerodynamics, simulation, wind tunnels, stress tests, and flight testing. Drug discovery is similar. A plausible molecule is only a starting point. The real challenge is predicting whether it will survive the experimental and physical tests imposed by biology along the path of discovery, pre-clinical studies, and human trials.
Physical AI operates on a similar logic. It is built to predict how real systems behave under chemical, biological, and time constraints. In drug discovery, that means predicting molecular behavior: binding, motion, solubility, permeability, metabolism, toxicity, exposure, pharmacology, and more.
This matters because drug discovery is fundamentally about how molecules behave inside and outside the cell. A molecule does not become a medicine because it is an interesting idea, or even because it looks plausible in a model. It becomes a medicine only if it behaves correctly in the crowded, complex, dynamic environment of cells that give rise to life. Predicting that behavior is the central challenge.
Different problems will require different forms of prediction: synthesis, binding, solubility, permeability, metabolism, toxicity, exposure, and pharmacology, each of which imposes different physical constraints and require different data. To succeed, physical AI will not use one predictive model. It will be a system of task-specific models, simulations, experimental feedback loops, and decision tools.
Molecular representation matters, too. Many current AI systems still treat molecules as strings or flat diagrams. But in drug discovery, molecules are 3-dimensional entities that change shape and properties within the crowded and changing biological environment. Small changes can have enormous consequences. Move one atom, change one bond, or alter one geometric arrangement, and a molecule can bind differently, distribute in the body differently, cause toxic effects, or fail entirely. Molecules are not static sketches. They are dynamic physical objects defined by electron clouds, geometry, motion, and quantum-mechanical interactions with their surroundings. Better prediction requires models that treat molecules much closer to that reality.
The practical lesson is simple: molecule generation is not molecule optimization. The real value in drug discovery comes not from proposing more molecules, but from learning faster which ones are actually worth making, and why. This is not just a scientific distinction. It is a practical framework for deciding where to invest capital, talent, and attention. The relevant question is no longer whether a company uses AI. It is which decisions its AI improves, how much those decisions improve, and whether those decisions ultimately lead to better medicines.
In other words, the key distinction is not between companies that use AI and companies that do not. It is between AI that helps scientists navigate existing information and AI that helps them predict behavior of novel molecules. Both matter. The first can be helpful is speeding up existing processes. But only the second can improve the odds that a molecule will become a medicine.
Real progress in AI for drug discovery should not be judged only by whether it makes teams move faster. It should be judged by whether it creates scientific leverage: better decisions about what to make, what to test, what to stop, and why.
A useful rule of thumb: be skeptical when a system is easy to demo but hard to validate. In drug discovery, the most valuable models are often not the ones that produce the flashiest outputs, but the ones that quietly help teams avoid bad bets and make better molecules.
Language-based AI will absolutely be part of the future of drug discovery. It will help search the literature, map the competitive landscape, write code, coordinate teams, and increasingly orchestrate closed-loop laboratories that connect prediction, experiment, and decision-making. It will make discovery organizations more nimble, better informed, and better coordinated.
Better prediction will also require better infrastructure. Progress in drug discovery depends not just on clever models, but on rigorous benchmarks, blind challenges, and the ability to scale. It also depends on infrastructure that is open, extensible, and designed for others to build on, much like shared platforms in the technology world have accelerated innovation far beyond what any one company could do alone. Not everything needs to be open, but not everything can be built from scratch within a single organization.
One especially promising area is co-folding, where the field has made striking progress in predicting biomolecular structure — progress significant enough to be recognized with the 2024 Nobel Prize in Chemistry. But structure alone is not enough. The next step is to connect those models to molecular recognition, energetics, mechanism, and design — the things that determine whether a structure is merely interesting or actually useful for making medicines.
This is the premise behind work done with my academic collaborators, colleagues at PsiThera, and the OpenFold community on “Grand Challenges for Predictive Modeling in Small Molecule Drug Discovery.” The impact of AI will ultimately depend on whether it helps teams predict which molecules to make next, which risks to prioritize, and which compounds have a realistic path to becoming drugs.
For investors and managers, that distinction matters. It separates operational efficiency from scientific leverage. The most important question is not whether a company can generate plausible outputs, but whether its technology improves real molecular decisions in ways that change asset quality, cycle time, capital efficiency, and ultimately probability of success.
That is the standard the field should reward: prospective validation over retrospective storytelling; physically grounded prediction over pattern matching alone; and open, extensible infrastructure over black-box claims that are difficult to test. If drug discovery makes that shift, AI will not merely accelerate existing processes. It will change how new medicines are found.
Woody Sherman, Ph.D., is the chief innovation officer of PsiThera and Chair of the OpenFold Consortium Executive Committee
Becky Pferdehirt is today’s guest on The Long Run.
Becky Pferdehirt, CEO, Radial
Becky is the CEO of Radial, the Emeryville, CA-based life sciences division of the Astera Institute. The nonprofit made news in March when it committed $500 million to Radial with a mission to reimagine data generation, sharing, infrastructure and tools necessary to create the best conditions for AI to help solve biological problems. Becky called it “unglamorous, unsexy infrastructure and tools” in an interview with STAT.
Becky comes to this work with a wide range of experiences in science. She started in academia, moved to industry labs, and eventually migrated to the business side of the biotech industry. She was most recently on the investing team at Andreesen Horowitz’s Bio & Health Fund.
This episode is about a fundamental question that a lot of people are wrestling with – how to lay down the ideal conditions for biology to thrive in the age of AI.
The Long Run is sponsored by:
AI has advanced from pilot programs into real biopharma workflows. But has it lived up to the hype? Drawing from an expert call with Dr. Paul Agapow, former director of data science at GSK, AlphaSense has a new report that dissects whether AI is on track to make drug development faster, cheaper, and more successful. Learn how AI is reshaping drug development in real time, turning AI execution into a key competitive differentiator.
Bioanalysis should not rely on which scientists at the CRO you get. These are assays, not consulting. Your results should be the same, regardless of who does the work.
“We love working with our CRO… when we get the right PI” is a red flag, not a compliment. And if the quality of your bioanalysis depends on which principal investigator happens to be assigned to your study, you don’t have a CRO partnership, you have a lottery ticket.
Dash was built around automation and standardized workflows specifically to eliminate that variability. Every study gets the same rigor and the same rapid turnaround, days not weeks, across ELISA/MSD, LC-MS, and PCR. GLP-compliant, transparent pricing, guaranteed outcomes. You shouldn’t have to ask who’s running your study. You should just get great data.
Visit www.dash.bio
Now please enjoy this conversation with Becky Pferdehirt on The Long Run.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
David Shaywitz
Back by popular demand – my highly curated list of summer content recommendations for TR readers.
Inside the Box – David Epstein
By now, you’ve read, or should have read Range, David Epstein’s previous book and his paean to generalists (my reflections here). Now, he introduces us to an intriguing paradox – the unexpected power of constraints to inspire greater creativity and productivity. It’s his most Gladwellian, for better (an intriguing read) and worse (conspicuously tidy at times), but it will absolutely keep you thinking long after you’ve finished. Bonus feature: the audiobook is read by the author, which I always prefer and applaud.
AI for Good – Josh Tyrangiel
I reviewed this book for the WSJ, and discussed in a recent TR, but it’s worth flagging again here because of its savvy focus on implementation challenges, and the many human-related barriers aspiring AI champions need to overcome.
A Giant Leap – Robert Wachter
Same story: I also reviewed this for the WSJ, and cited in passing in several TR columns (eg here). The book deserves highlighting because of the unusually good job Dr. Wachter does in contextualizing not only AI but technology more generally within the complex process of healthcare delivery.
The Future of Seeing – Daniel Sodickson
Dr. Daniel Sodickson, the author of this wonderful account of the history of imaging, is (disclosure) a med school classmate and friend, and his deep knowledge, visceral enthusiasm, and wry humor practically leap from every page. He delivers a fascinating look at the past, present, and possible future of extraordinary imaging technologies – as well as a warm introduction to the people responsible.
The Way of Excellence – Brad Stulberg
Brad Stulberg, a prominent executive coach and popular writer and podcast host (“Excellence, Actually”) has written an appealing guide to performance excellence, one that eschews quick fixes in favor of sustained engagement. While shading a bit aphoristic and “coachy” for my taste, the underlying substance is remarkably solid and informed by evidence; there’s a real “there” there.
Outlive – Peter Attia
This book, by Dr. Peter Attia (whose reputation was tarnished this year after his association with Jeffrey Epstein was revealed) is on the list more than a year after we discussed the book in TR. Why? Because while many physician-authors have written knowledgably about healthy living, Attia’s book seems to have motivated – truly lit a fire under — far more people (including more physicians) than most other similar books I’ve read, including many that (in my view) are scientifically on target with their shared lifestyle message of eat thoughtfully, move joyfully, and connect with friends and community. Somehow, in Attia’s hands, the message seems more compelling. I clearly don’t share his obsession with relentless optimization, but his book, together with some lessons from (David!) Epstein on constraints, make me wonder if there’s some real value in offering at least a modicum of gratuitous specificity.
Two recommendations here, both available without charge: the always rigorous Ground Truths, from Dr. Eric Topol, and the recently-launched Substance Over Noise, from Dr. Michael Albert.
Ground Truths: Eric Topol spends quality time with science newsmakers; he’s apparently become a mandatory stop for authors who’ve just launched a book. The happy result: consistently interesting and timely discussions.
Recommended recent episode: Topol’s chat with Sebastian Mallaby author of The Infinity Machine.
The Future of Everything: Stanford’s Russ Altman speaks with leading scholars (generally from Stanford) at the cutting edge of a range of fields, making for especially thoughtful discussions.
Recommended recent episode: Altman’s conversation with communications professor Angèle Christin on influencers.
Business History: I’ve highlighted this show before, and it’s remained one of my favorites – tight, engaging stories about (yes), the history of a particular business.
Recommended recent episode: history of Veuve Clicquot champagne.
Acquired: Also previously recommended, and also still good – very deep dives into specific businesses.
Recommended recent episode: John Bogle and Vanguard – “the most effective vehicle ever created for participating in the fruits of American capitalism.”
Approved: This is a highly recommended, relatively new podcast hosted by Alex Kesin and Matthew Pech. Essentially “Acquired” but for FDA-approved drugs, the show features deep dives into the history of how the medicine came to be.
Recommended recent episode: history of Keytruda – a captivating story I’ve discussed in detail here (also referenced by the show).
The Happiness Lab: A return appearance by a perennial favorite. Hosted by Yale professor Laurie Santos, the show features thoughtful, accessible interviews with experts around topics in positive psychology.
Recommended recent episode: interview with Shawn Achor, author (most recently) of The Power of Beliefs.
NEJM-AI Grand Rounds: A podcast hosted by Raj Manrai and Andrew Beam, featuring discussions with leaders at the intersection of AI and medicine. I’ve highlighted the show before, and it remains excellent.
Recommended recent episode: fascinating discussion w Travis Zack, the CMO of Open Evidence (my TR on Open Evidence here).
Of course, TR readers will continue to enjoy and appreciate Luke’s always captivating interviews with biotech leaders on The Long Run.
Finally: have a great summer! Savor the opportunity to move outside, connect with family and friends, enjoy late dinners in the delightfully warm air. Find a few minutes to reflect on what’s most important to you, and pursue meaningful engagement – with people, ideas, interests, communities.
Be kind and be well.
– David
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Jeremy Levin is today’s guest on The Long Run.
He is the executive chairman and founder of New York-based Ovid Therapeutics. He’s also a board member of the Biotechnology Innovation Organization, and a past chairman of the BIO board of directors.
Jeremy Levin, executive chair, Ovid Therapeutics
Jeremy has a new book out today, called “Biotech in the Balance: Saving a Strategic Industry in an Age of Distrust.” It’s for biotech executives, policymakers, or any curious citizen interested in how we better harness biotechnology for the betterment of the maximum number of people around the world.
This book is not just a laundry list of what’s wrong with the industry or the public. It contains 10 commitments. These are practical suggestions of things companies can do to make for a healthier industry, and to help create a more stable, thriving industry that earns more public support.
I hope you enjoy this conversation, and that it might help serve as a conversation starter in your circles.
The Long Run is sponsored by:
AlphaSense is an AI-powered market intelligence platform designed to help investors and operators find critical insights faster. Think of it as a search engine for the entire financial and life sciences landscape, aggregating over 500 million documents across company filings, clinical trial registries, FDA documents, earnings calls, expert transcripts, broker research, and industry publications into one unified, searchable platform.
For those researching biotech, medtech, and the broader life sciences landscape, it is a game-changer, purpose-built to surface the clinical, regulatory, and commercial insights that matter most. You can use Smart Summaries to instantly synthesize thousands of pages of trial data, regulatory filings, and KOL commentary, while Channel Checks and expert insights help you understand how therapies are actually being adopted in the real world across physicians, payers, and hospital systems, not just how they are modeled on paper.
Instead of spending hours digging through 10-Ks, AlphaSense surfaces the ‘Alpha’—the specific insights that move the needle—in seconds. To experience it firsthand, you can start a free trial at alpha-sense.com/thelongrun.
Bioanalysis should not rely on which scientists at the CRO you get. These are assays, not consulting. Your results should be the same, regardless of who does the work.
“We love working with our CRO… when we get the right PI” is a red flag, not a compliment. And if the quality of your bioanalysis depends on which principal investigator happens to be assigned to your study, you don’t have a CRO partnership, you have a lottery ticket.
Dash was built around automation and standardized workflows specifically to eliminate that variability. Every study gets the same rigor and the same rapid turnaround, days not weeks, across ELISA/MSD, LC-MS, and PCR. GLP-compliant, transparent pricing, guaranteed outcomes. You shouldn’t have to ask who’s running your study. You should just get great data.
Visit www.dash.bio
Now please enjoy this conversation with Jeremy Levin on The Long Run.
David Shaywitz
This week’s Duke-Margolis Health Policy Conference (video here) left me feeling there’s a remarkable opportunity to leverage technology to dramatically improve healthcare — and there’s almost no chance of it happening within the current healthcare system.
A number of speakers (including UCSF’s Bob Wachter, my WSJ review of his recent book here) were genuinely enthusiastic about the transformative possibility of AI, a technology they said made them feel more optimistic about the opportunity to positively impact healthcare than they would have ever thought possible.
They described opportunities to organize disparate data, enhance chronic disease management, and provide patients with access to increasingly useful medical information and guidance.
Yet — the leading way hospitals are apparently actually leveraging AI?
Upcoding and other approaches to maximize billing.
As Jacob Shiff, Chief AI & Technology Officer, CMS Innovation Center, explained,
“Fundamentally, technology is a tool. People use tools based on their incentives. And the dominant incentives in our healthcare environment today are to maximize billable activity volume and intensity. And that is where we are seeing most technologies being deployed. A tool like an AI scribe can enable a clinician to shift non-reimbursable activities towards reimbursable activities and maximally capture every billable encounter, every billable event in an encounter. That’s a path we’re on. It’s going to increase health care costs. It may or may not have the value to show for it.”
And payors, of course, are leveraging AI in kind.
The result is what appears to be a fatal mutual embrace, a death spiral for medicine, where most everyone involved in the current system of care appreciates how misaligned the incentives are, yet everyone continues to essentially do what they’re already doing, only more so — and the misery just gets worse.
Costs remain high and the experience of both patients and providers seems to be degrading at an astonishing pace.
It’s not surprising to me that more physicians are opting out of the system, that more patients are turning to consumer health and wellness options, and that savvy entrepreneurs are trying to construct approaches to health that build from the consumer side, as Whoop and Oura (among others) seem intent on doing.
Of course, it’s easy and appropriate to critique (as I and others such as Eric Topol have) aspects of consumer health, like the embrace of tests (such as biological age) that aren’t validated for individuals.
But take a half step back and compare the trajectories of consumer health companies with the traditional health system.
Consumer health companies tend to offer elegant patient-centered design, often an emphasis on patient-ownership and management of their data (eg Superpower), and a proactive mindset.
In contrast, traditional healthcare has effectively locked down patient data (as several conference speakers commented), is entirely oriented around reimbursement (or denying reimbursement), and remains fixated on documenting adherence to defined process metrics vs actual patient need.
While it’s clear the guilds of healthcare will fight fiercely to prevent these tech-enabled interlopers from intruding on their turf, popular dissatisfaction with the status quo seems likely to encourage politicians to at least offer disruptive tech-enabled startups meaningfully more room to operate, which will likely only increase as the benefits become more obvious and are experienced by more people.
Moreover, efforts by the health system to appeal to a gauzy, idealized view of medical care and insist we must defend it will carry less weight as such images feel ever more remote from our lived experiences and interactions with today’s health system.
It’s not only about AI, to be sure, but AI is a powerful driving force. Today’s AI models, University of Pennsylvania’s Ethan Mollick reminds us, are “the worst they will ever be,” and the technology is only getting better.
No one would say that about the trajectory of our current healthcare system.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
David Shaywitz
AI is here – everywhere it seems. How are we doing on translating this extraordinary promise into palpable value?
As utopian AI “Accelerationists” have battled catastrophizing “Doomers” over competing visions of which eschaton is likely to be immanentized, a less visible but perhaps more consequential constituency has quietly focused on applying the still-evolving technology to the intractable problems of the present.
These doers — understated, practical and ridiculously stubborn — are the subject of a compelling new book by Josh Tyrangiel, AI for Good, I’ve recently reviewed in the WSJ. (Sadly, my original lede — the one that starts this section — didn’t survive the editorial scalpel.)
Tyrangiel’s key points will all sound familiar to readers of this column: AI implementation is not plug and play — it requires deliberate effort and domain expertise. Critical players include lead users on the organization size focused on actual problems to solve, as well as a capable tech partner with a high EQ translator who can inhabit the problems the organization is trying to solve and translate that into technology spec for developers.
Even with all these in place, getting new technology to do what you want is inherently challenging, as a very well-intentioned collaboration between OpenAI and Khan Academy reveals, and is often further complicated by organizational antibodies, which aspiring innovators need to manage.
Approaches to this range from racing to take advantage of the temporary loosening of bureaucratic hurdles due to a crisis (Gen Perna of Operation Warp Speed) to working under the radar like former IRS Commissioner Danny Werfel, who advanced his agency’s AI capabilities under the cover of its reputation as “slow and boring and technologically hopeless.”
Still, resistance is everywhere, and medicine earns a particular callout. “Hospitals,” Tyrangiel observes, “may have exalted missions, but they’re just as full of territorial jerks as investment banks.” Several examples from the Cleveland Clinic — sure to resonate with innovators in academic medical centers everywhere — bring the point home.
A second AI book I reviewed in the Journal, I Am Not A Robot, by Joanna Stern, looks at how AI is, and is likely (in the near future) to impact the lives of individuals. Again, the issue isn’t so much the technology itself as how it’s used. AI to improve early cancer detection in mammography: good; AI to use colorful overlays to upsell patients on marginal dental procedures: less good. Household robots: not ready for prime time –- although Stern reports AI massage robots deliver “some great ass work.”
Minds and hearts, naturally, are of particular interest, as both learning and love involve discovery & struggle, negotiation & compromise -– and Stern counsels that we should be wary of AI bearing effortless solutions.
The application of AI to biopharma was recently discussed with characteristic thoughtfulness by veteran Journal reporter Peter Loftus, drawing conclusions that, again, will not surprise regular readers of this column.
We should separate out some highly questionable claims, such as the suggestion that Takeda’s TYK2, acquired from Nimbus, was discovered using AI; as physician-scientist and healthtech investor Patrick Malone noted three years ago, the medicine “was discovered using a type of computer-aided drug discovery called structure-based drug design” which qualifies as AI to the extent that “AI is a catch-all term for methods that use computers to solve a task.”
What seems most clear is that the lion’s share of meaningful progress has been in operations; as one Lilly executive notes, “The reality is where we’ve seen all the benefits in AI so far is not actually in drug discovery — it’s actually in the rest of the process.”
As readers of this column recognize, this is what several pharma leaders – including the CEOs of Lilly and Novartis – have acknowledged for years.
To be sure, many of the same companies that were previously persuaded by management consultants that the secret to radically improved R&D efficiency involved building and then posing in front of an enormous data dashboard have now apparently been convinced that R&D deliverance (or at least an opportune distraction) requires constructing a ginormous supercomputer that goes up to 11 (or to 12, in the case of Roche’s, evidently).
More substantively, key challenges limiting the impact of AI in R&D, as Andreas Bender and colleagues, including Jack Scannell and me, describe in a forthcoming publication include:
Data quality and character. Biological data has three properties that limit AI’s impact: scarcity (few endpoints that truly matter vs. abundant proxy data), fragility (labels like “works” or “binds” are highly conditional, not categorical), and epistemic opacity (biological complexity limits the line of sight between proxy measures and outcomes we actually care about).
Confusing map and territory. Models that look promising in journals often function poorly in actual R&D contexts — a problem described compellingly in a recent Nature Medicine publication by Azad, Krumholz, and Saria in the related context of clinical care.
There’s a fascinating new proposal by two authors familiar to TR readers: Ziad Obermeyer, an emergency room physician and health service researcher (profiled in TR here), and Katherine Baicker, a health economics scholar and Provost of the University of Chicago, also the co-author of a prominent recent study debunking a range of approaches to workplace wellness (discussed in TR here).
In their new paper, the authors start by acknowledging the urgent need for surrogate endpoints for clinical trials, essentially to accelerate knowledge turns. TR readers obviously get this and wrestle with this every day. The issue, of course, has been exactly the “epistemic opacity” discussed above –- we generally don’t understand as well as we need to the connection between what we can measure and the endpoints about which we care.
Consequently, when we suggest seemingly sensible surrogate endpoints, they often fail. Obermeyer and Baicker bring the painful receipts, captured by Obermeyer in this post (part of an excellent thread on the topic):
Instead of using our flawed causal understanding to select a candidate marker, the authors suggest that AI models trained on comprehensive, readily available physiological data (“ECGs, imaging, laboratory results, and clinical notes”) can generate a “surrogate index” –an empirical, composite, predicted probability of long-term clinical outcomes that shifts proportionally to a drug’s true treatment effect.
It’s an approach that reminds me of polygenic risk scores (this is a useful PRS explainer from Eric Topol at Ground Truths), only more dynamic –- and seems extremely promising, albeit with a number of caveats, as the authors highlight.
A particularly important challenge –- with which I’ve engaged since my very first pharma role in experimental medicine at Merck two decades ago – is that the pragmatic (vs academic) utility of a surrogate marker is having sufficient confidence in it to inflect the direction of a drug development program (a surrogate marker that actually counts as an acceptable proxy by the FDA for a traditional endpoint is an even higher bar).
Essentially, you need to be able to kill a beloved program based on the results of the surrogate marker in early studies –- and for all sorts of organizational reasons, this is brutally difficult to achieve. But if AI-derived markers earn a reputation for being sufficiently robust –- they could be embraced as the authors propose, and meaningfully accelerate development, and (beyond what the authors write) perhaps find acceptance (at least provisionally) as an endpoint acceptable to the FDA.
A recent Ground Truths post by Eric Topol describes a fascinating — and concerning — dichotomy between the inexplicably slow general adoption of task-specific deep learning, which has proven its value in healthcare with some rigor, and the remarkably rapid general adoption of generative AI, a newer and more versatile generation of deep learning that still urgently needs more rigorous validation.
The need for more robust evidence around genAI aligns with a just-published Nature Medicine editorial, “Show us the evidence for the value of medical AI,” as well as a plea from leading practitioners such as Raj Manrai of Harvard’s Department of Biomedical Informatics (my home department).
Manrai writes that we need “[p]rospective clinical trials. Health systems investing in infrastructure now. Monitoring frameworks that track not just diagnostic accuracy but safety, efficiency, and cost.” He adds, “The science has reached a point where trials are justified.”
Topol’s strongest examples of the delayed implementation of deep learning applications come from imaging. The technology can extract clinically relevant information from scans obtained for other reasons: cardiovascular risk from chest X-rays or mammograms, osteoporosis risk from both chest films and retina images(!), body composition and metabolic risk from CT or MRI, liver disease from echocardiograms, and systemic disease risk again from retinal images (as Topol points out, the retina appears to be an unusually rich source of signal detectable by AI).
This is the promise of opportunistic AI: making better use of data medicine already collects, without additional radiation or inconvenience, and potentially at minimal additional cost.
The key implementation question is how to turn that additional signal into better care. Some outputs — calibrated risk estimates, bone-density measures, body-composition phenotypes — may fit naturally into established prevention pathways. Others, especially potential cancers, nodules, or occult lesions, raise the familiar screening concerns: false positives, downstream testing, overdiagnosis, anxiety, and uncertain net benefit.
The real opportunity is to convert AI-derived signals into clinically useful intelligence, echoing Zak Kohane’s aspirational description of medicine as “fundamentally an information- and data-processing discipline.”
Adds Kohane, the Chair of Harvard’s Department of Biomedical Informatics, “How you distill data into knowledge, and how you take that knowledge and put it in practice, is at the heart of medical science today.”
Several colleagues and I have been thinking about how to make AI-enabled monitoring more context-aware and clinically useful; a related perspective is currently under peer review.
Biopharma and healthcare, like the rest of the world, are actively engaged with the challenge of figuring out where and how AI can be usefully applied. As “AI for Good” shows, implementation is difficult and requires deliberate work and the right people. We should use it deliberately and strategically, rather than embrace it reflexively or mindlessly; we also shouldn’t underestimate its exceptional promise, even as we foreground the significant concerns inherent in its development. In pharma, using AI to accelerate discovery is challenging and so far has arguably been more succès d’estime –- and succès de tech VC dollars – than transformative progress. Meanwhile, AI-enabled surrogate indices represent a promising (but still unproven) approach to accelerating clinical trials. Evidence suggests the application of deep learning to medical imaging can provide important insights, value that has largely been overlooked by traditional medicine, and a market opportunity thoughtful entrepreneurs are sure to recognize.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Richard Pops is the latest guest on The Long Run.
Richard is the chairman and CEO of Alkermes. He’s one of the longest-serving biotech leaders ever, with a 35-year career at the helm of a single company. Rich is also one of the more issue-oriented and politically engaged industry leaders. As the head of a mid-sized biopharma company, he has an unusual perspective as the only member of the board for both the Biotechnology Innovation Organization and PhRMA – the trade groups that represent small and large biopharmaceutical companies.
Richard Pops, chairman and CEO, Alkermes
This summer, Rich is stepping down as CEO but will remain chairman of the board of Alkermes. The company has had its ups and downs, even a battle with activist investor, but it’s in as good a position now as ever. It’s profitable, has $1.3 billion in cash, and is executing on a late-stage clinical development plan for what could be its biggest product ever.
Alkermes – along with Takeda Pharmaceutical and Eli Lilly through its acquisition of Centessa Pharmaceuticals – are blazing a trail in a new class of medicines for sleep disorders. These drugs are orexin receptor agonists.
Orexins are neuropeptides in the hypothalamus that are considered “master conductors” that regulate wakefulness, arousal, and energy homeostasis, acting as key stabilizers of the sleep/wake cycle. The initial application for orexin receptor agonists is in narcolepsy, aka excessive daytime sleepiness, but there’s more. The biology is fascinating and generates all sorts of hypotheses about whether these orexin drugs will be useful for fatigue associated with multiple sclerosis and Parkinson’s, addiction, depression and other neurological conditions.
In this wide-ranging conversation, Rich reflects on his long history at the company and in biotech, and how Alkermes evolved to seize this opportunity in orexin biology for narcolepsy and many more neurological applications. We also discuss this moment in biotech writ large, brimming with possibilities and with plenty of stormy weather.
The Long Run is sponsored by:
In Life Sciences, the ‘information gap’ is where portfolios go to die. You aren’t just tracking tickers; you’re tracking Phase III data, FDA shifts, and patent moats. If you’re waiting for the news cycle to catch up, you’re already too late.
That is why I use AlphaSense.
AlphaSense is an AI-powered market intelligence platform designed to help investors and operators find critical insights faster. Think of it as a search engine for the entire financial and life sciences landscape, aggregating over 500 million documents across company filings, clinical trial registries, FDA documents, earnings calls, expert transcripts, broker research, and industry publications into one unified, searchable platform.
Instead of spending hours digging through 10-Ks, AlphaSense surfaces the ‘Alpha’—the specific insights that move the needle—in seconds.
To experience it firsthand, you can start a free trial at alpha-sense.com/thelongrun.
Bioanalysis should not rely on which scientists at the CRO you get. These are assays, not consulting. Your results should be the same, regardless of who does the work.
“We love working with our CRO… when we get the right PI” is a red flag, not a compliment. And if the quality of your bioanalysis depends on which principal investigator happens to be assigned to your study, you don’t have a CRO partnership, you have a lottery ticket.
Dash was built around automation and standardized workflows specifically to eliminate that variability. Every study gets the same rigor and the same rapid turnaround, days not weeks, across ELISA/MSD, LC-MS, and PCR. GLP-compliant, transparent pricing, guaranteed outcomes.
You shouldn’t have to ask who’s running your study. You should just get great data.
Visit www.dash.bio
Now please enjoy this conversation with Richard Pops on The Long Run.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
Martin Burke is today’s guest on The Long Run.
Marty is a professor of chemistry, Director of the Molecule Maker Lab at the University of Illinois at Urbana-Champaign. He’s well known in the chemistry field for his pioneering work on blocc chemistry: iterative carbon-carbon bond formation that makes chemical synthesis faster, highly automated, and user-friendly.
Martin Burke, professor of chemistry, University of Illinois; co-founder, Excelsior Sciences
That was a big deal for drug discovery a decade ago when it was published in Science. It’s an even bigger deal today as companies feel it’s imperative to automate physical steps in drug discovery and remove process bottlenecks so that AI can fulfill its potential.
Marty is eager to see his academic work translated into products that help people. He’s overseen work on small molecules that treat fungal infections without causing toxicity to the kidneys and ones that mobilize iron transport through tissues to treat anemia in various forms. Most recently, he’s been working on a startup in New York, Excelsior Sciences, that has raised $95 million to build an automated platform to quickly synthesize small molecules and further test and optimize them with AI. The platform is designed from the ground up so that the small molecule compounds are like tokens that AI can learn from – like large language models already with text.
Marty brings an infectious enthusiasm to his work. I enjoyed this conversation and I think you will, too.
The Long Run is sponsored by:
It’s earnings season, which means investors are deluged with financial reports from one company after another. It’s hard to absorb the tidal wave of information, and find enough time to think about what it means.
To help people better prepare for earnings season, AlphaSense recently hosted a webinar with a pair of investors and a former Novartis competitive intelligence director. AlphaSense’s AI platform draws on a vast pool of regulatory filings, analyst reports, expert interview transcripts, and more.
Register for the “AI Playbook for Earnings Prep and Analysis” and watch the replay here.
Bioanalysis should not rely on which scientists at the CRO you get. These are assays, not consulting. Your results should be the same, regardless of who does the work.
“We love working with our CRO… when we get the right PI” is a red flag, not a compliment. And if the quality of your bioanalysis depends on which principal investigator happens to be assigned to your study, you don’t have a CRO partnership, you have a lottery ticket.
Dash was built around automation and standardized workflows specifically to eliminate that variability. Every study gets the same rigor and the same rapid turnaround, days not weeks, across ELISA/MSD, LC-MS, and PCR. GLP-compliant, transparent pricing, guaranteed outcomes.
You shouldn’t have to ask who’s running your study. You should just get great data.
Visit www.dash.bio
Now please enjoy this conversation with Marty Burke on The Long Run.
Please subscribe and tell your friends why it’s worthwhile. Quality journalism costs money. When you subscribe to Timmerman Report at $199 per year, you reward quality independent biotech reporting, and encourage more.
