Transcription
Westin: This is not yet another story about the promise of artificial intelligence. It's the first in a series of stories that go beyond the hope and the hype to see where AI is making a real difference today, starting with something important to us all, our medical care.
-Good morning.
-Good morning.
Westin: On any given day, Tennessee-based oncologist Samyukta Mullangi could see as many as 25 patients.
-Medical oncologists like me are pressed for time and we're overwhelmed with sort of clinical duties. Trying to do that work of charting in ahead of time, seeing the patient, documenting after the fact, and also now trying to like squeeze in record retrieval in the middle of everything is just very impossible. And so you have the concept of pajama time where oncologists and other physicians are just sort of finishing up their daily work at home after, you know, after the kids are put to bed and after dinner and it just contributes to a lot of provider burnout.
Westin: Dr. Mullangi is not alone. All across the country, doctors in all specialties are often stretched thin. The American Medical Association found that almost half of U.S. positions experience at least one symptom of burnout. It's little wonder that our doctors are feeling the burden, given the explosion of medical research. So Daniel Nadler decided to do something about it.
-The rate of doubling in... of medical knowledge in 1950 was roughly every 50 years. In 2025, there are different estimates and there are different numbers. In a study in the British Medical Journal and another study in Nature, they found that the rate of doubling of medical knowledge was 73 days.
Westin: Nadler's PhD thesis from Harvard was on analyzing derivatives, which he turned into a startup using machine learning for financial analysis. After selling that company for half a billion dollars, he turned his attention to helping doctors make sense of the tsunami of medical research coming their way.
-We looked at this and we did another analysis and we said, "Let's ignore the doubling for a second and just ask the question, if you had to read just the top third of peer-reviewed medical literature just within your specialty, which is not ideal, right? That means no cardiologist is reading anything in neurology and vice versa, which is not ideal. But even if you just said that, how long would it take every day for a specialist to just read the top third of peer-reviewed medical literature just within their specialty? And the answer turned out to be something about... something like 9 hours. So, practically, that's obviously impossible. They would never see patients or never see their family or they would never sleep.
Westin: And that led Nadler to found Open Evidence in 2022.
-Open Evidence is designed to do for physicians what the advent of computer systems, let's call it that, on Wall Street achieved for Wall Street knowledge workers.
Westin: Open Evidence is an AI model trained on medical literature. Carefully curated to ensure high-quality results.
-For the first time in the last, let's call it 3, 4, 5 years maximum, we've reached a point in the sophistication of artificial intelligence, of computers broadly, that they can store not just the right letters and the right words in the right order, but they can understand the semantic meaning of the findings of these studies. So Open Evidence is an artificial intelligence, it's a computer system that's able to understand the semantic meaning of the findings of these studies so that it can act as a brain extender to physicians who have, even in the best and most generous reading of what they have to do as a physician in terms of keeping up with the pace of medical knowledge, such that they don't have to spend 9 hours a day just reading the top third of peer-reviewed medical journals.
Westin: Doctors are piling into the platform, which says it has signed up around 50% of all doctors in America and it's adding 65,000 every month. Investors are piling in too. Its latest founding round valued the company at 3.5 billion dollars. What is it that Open Evidence should be relied upon to do and what do we still need the physician to do? For example, can Open Evidence diagnose?
-No. So the physician is still relied upon to do everything that a physician was always relied upon to do. I see Open Evidence as a continuum or a continuation of a very traditional technology called search, right? So, historically, physicians needed to search for findings in medical journals. That's not a new behavior. They've been doing that for years and years and years. One way to think about this is we spend a lot of time as a society celebrating the golden age of biotechnology, and we should, right? Every metric and proxy you look at in the data shows that we are accelerating the rate of drug discovery, including using artificial intelligence, we're accelerating the rate of drug development, and so we celebrate that. We're in this golden age, and it's amazing, right? This golden age of biotechnology. But what's not talked about a lot, or enough, is that this golden age of biotechnology is really the dark ages for physicians in terms of burnout. We expect that almost all physicians in the United States will be on the platform within the next year.
Westin: Dr. David Reich is President and Chief Clinical Officer for Mount Sinai Health System in New York. He uses Open Evidence, but as part of a larger range of AI models they are integrating into their hospitals.
-I have the app on my phone and it's available through our medical school library and people do use it. However, we do also work with Chat GPT and they've created an environment where we can have our medical students ask questions that contain protected health information. And that protected health information stays within the cyber secure environment that we work so hard to maintain. And so I think cyber security remains a key consideration in any tools that are a great assistance to us and I'm sure Open Evidence will work very hard to address that along with others. But I'm very enthusiastic about it, I think it's a great tool.
Westin: One of the reasons Open Evidence has become so popular with doctors so quickly is that it licenses the best medical literature from trusted sources and it's free to doctors relying on advertising for its revenue.
-We, right from the start, licensed content, licensed journals, licensed information, medical information from the relevant copyright holders. So we have an agreement with the Massachusetts Medical Society, which owns the New England Journal of Medicine. It's a non-profit. We have a licensing agreement with the American Medical Association. Again, it's a non-profit, but it owns the The Journal of the American Medical Association, as well as all the specialty journals, JAMA Oncology, JAMA Neurology, and so on and so on and so on, not just those. So we took a very different approach to copyright, to licensing, to all of this, and we're extremely proud of the fact that we're probably the only pure AI company in America that is not currently being sued for copyright violations. Our business model is the exact same as Google, which is, I think it's public, Google is one of our largest investors and has been an enormous patron to the company in many ways. So you can build very successful software or service companies and there are many examples of 10, 20, 30, 40, 50 hundred billion dollar market cap software or service companies. But once you start getting into the rare area of multi trillion market cap companies, it's notable that almost all of them again, excepting NVIDIA and Apple, are either primarily advertising business models or have advertising as a significant component of what they do. And I think the explanation there is it is the business model that most aligns the incentives of the platform and the users to make sure you're delivering to your users the highest quality product possible because your incentive is not reducing cost. Your incentive is attracting more users and increasing your engagement.
Westin: AI tools like Open Evidence are already providing much needed help to physicians who must focus on their patients even as they need to keep up with the deluge of new research. But for AI to realize its full potential in healthcare Dr. Reich says it needs to become fully integrated into the workflow.
-Mount Sinai is doing the genetic information analysis on up to a million patients in partnership with Regeneron and we're several hundred thousand patients into this and we have a vast trove of information on patient medical images and we have incredible information in our electronic health record. Now when we start to marry all of those data sources together and follow the promise of AI, in the not too distant future, I should be able to say to you when you come in, not only did I screen you and I found particular risks, not only do I have care pathways which suggest how I should go forward, but it's specific to you, to your family history, to your genetic markers and, hopefully, giving you the best and safest possible experience. So think of the future as being much more personalized and the advance of technology is so inspiring right now that I think that what I've witnessed over several decades of medicine could vastly change in the next several years as long as we learn the lessons of past mistakes of being maybe too exuberant about technology. And making sure that people who are truly in touch with that social contract between patients in this nation and the payers and the government and the providers that we actually find really good solutions.
Westin: While Dr. Reich emphasizes the integration of all the data into a single workflow, Daniel Nadler envisions a world connecting physicians with others around the world who are working on the same clinical challenges.
-As Open Evidence develops, if we go back to my metaphor of the sort of 1940s World War II telephone operator who's routing and connecting a human to another human on a battlefield, in this case, it might actually end up in a world where the AI is the least interesting part of the technology. And what's really happening is the AI is serving as connective tissue between a human and a human, between a human physician presenting some atypical combination of symptoms and another human somewhere in the country that is an expert on that and where the job of the AI is as far as possible from answering the question, and it's much more about getting out of the way as quickly as possible and connecting that one human to another human and that's a very wonderful and sort of optimistic vision for what the future of AI can be. You know, most scenarios for the future of AI, or many, are very dystopian. I can't comment on what happens outside of medicine, but in medicine I think you have a very beautiful possibility where the technology ends up serving as connective tissue between a human and a human.
Westin: But whether it's connecting the doctor with the data or the human with the human, right now AI products like Open Evidence are providing much needed relief for practicing physicians like Sam Mullangi.
-I would say that for me Open Evidence solves like two rather unrelated but maybe orthogonal problems. One is that actually Open Evidence taps into the entire medical corpus of academic literature, which actually, for the most part, tends to be paywalled. The second thing that it does, I think, is just sort of marrying the best of current AI capabilities around natural language processing, and reasoning to try to interpret my request, and retrieve records easily in a very time-efficient manner. So having sort of smart AI tooling that is able to... provide fast queries has just been really... been a game-changer for me.
Westin: And that is one application of artificial intelligence that is making a real difference in the here and now.