Subscribe to The Podcast by KevinMD. Watch on YouTube. Catch up on old episodes!

Rural doctors hit a ceiling around 35 patients a day, and hiring more clinicians will not move it. Tod Stillson, a family physician, medical device inventor, and health care entrepreneur, argues that the physician shortage is not a headcount problem but a knowledge-scaling problem, and that AI with a doctor in the loop is the only realistic way to extend a physician’s judgment to more patients without replacing the human relationship. This episode is based on his article “How artificial intelligence scales physician extension,” published on KevinMD. You will hear why he spent two years codifying his own clinical judgment into software, why his text-based triage system outperformed general-purpose AI in a recent study, and what physicians risk if direct-to-consumer companies keep capturing demand while doctors stay on the sidelines. You will also learn why governance of the medical knowledge base has to come from physicians, not from startups in San Francisco or Boston. Listen for a grounded case for AI as capacity multiplier, not replacement.

Partner with me on the KevinMD platform. With over three million monthly readers and half a million social media followers, I give you direct access to the doctors and patients who matter most. Whether you need a sponsored article, email campaign, video interview, or a spot right here on the podcast, I offer the trusted space your brand deserves to be heard. Let’s work together to tell your story.

PARTNER WITH KEVINMD → https://kevinmd.com/influencer

SUBSCRIBE TO THE PODCAST → https://www.kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://www.kevinmd.com/recommended

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast. Today we welcome back Tod Stillson, family physician, medical device inventor, and health care entrepreneur. Today’s KevinMD article is “How artificial intelligence scales physician extension.” Tod, welcome back to the show.

Tod Stillson: Kevin, it’s great to be with you. I love chatting with you about the host of things that are in physicians’ minds around the country. So we’re going to talk about one of those today.

Kevin Pho: All right, well, thank you again, and always interested specifically to talk AI with you. I know your last few episodes were always AI related, and it is disrupting us in so many different ways. What’s your latest perspective about it?

Tod Stillson: Yeah, my latest perspective, and this has to do with what I’m actually doing in real practice, in the marketplace of my on-demand urgent care called ChatRx. But at the end of the day, I think AI has a real potential to be a problem solver for us when it comes to how we manage extending our knowledge base.

You know, the reality is in health care, in the health care system, we’re not so much constrained by knowledge, Kevin, as we’re constrained by how many times that knowledge can be applied in a day. AI really changes that equation completely. As an example, in my real practice for 30 years, I’d see 35 patients a day, right? But that’s a ceiling. There is a capacity.

Kevin Pho: Like even that’s a lot. See, your listeners are going to go, 35 patients, that’s a whole bunch.

Tod Stillson: Yes, it was a whole bunch. But at the end of the day, there’s still a ceiling. I couldn’t add more people in. It could limit my knowledge, even if I wanted to share that knowledge with others. There was a capacity and constraint issue on that, and that’s becoming a big issue in our country with capacity and access and so forth.

Kevin Pho: So the fundamental issue is not knowledge. It’s how you get that knowledge from our training to the most people, right?

Tod Stillson: Absolutely. And you yourself know, in a rural area where resources are constrained, you’ve lived it firsthand.

Kevin Pho: I have. AI, you know, it’s disrupting, like I said, so many industries. But of course, if we use it the right way, or if we come at it from the right perspective, it could also help us. So specific to AI, how can that help scale our knowledge?

Tod Stillson: Yeah, so I think it requires us to build the systems that support the actual physician doing the work, right? So this is not coming at it from a pure business sense. It’s coming at it from a pure sort of, the physician wants to take care of the patient, who wants to share that knowledge with more people. This allows us, if you build the system using AI to support you with your codified clinical judgment or your decision-making assistance, it allows your knowledge base to be applied to more people and with greater capacity, but doing it in a constrained way.

The words that we use in this software development space for AI is a doctor in the loop, right? So a human or a doctor in the loop. So it’s not just AI run wild, it’s actually the physician inside orchestrating it to help us accomplish the care that we’re trying to accomplish, deliver care to patients. And again, that’s applying the knowledge to it.

So, Kevin, here’s the honest truth. It took me almost two years to take my knowledge base of 30 years of experience and feed it into a piece of software to sort of codify Dr. Stillson, so to speak, and allow the system to sort of function. By the way, it’s not just Dr. Stillson, it’s national guidelines. High-functioning rural doctor, we follow guidelines, OK? And so you are able to orchestrate that into a system that assists me then in delivering care to more people, right?

So go back, you know, and all the listeners right now, a lot of them have PAs, NPs, nurses, a lot of people around them. The rural, your rural listeners will go, yeah, that’s how it’s orchestrated. Nowadays, we’ve got 5, 6, 7 people around us, and you and those doctors are used to sort of a standing order sequence, a triage process. This is what we do every time for these patients, that the teams, they all kind of know how to orchestrate that. And to be honest, that works to some extent, but even that has ceilings to it. OK, it still does have ceilings to it.

But you can take that same engineered process, Kevin, and apply it to technology, i.e. AI, and allow it to assist us in engineering that knowledge base. And the end result is your knowledge can be scaled up even faster, quicker, and more applicable to a lot of people. And that’s what those of us in the technology sector with medicine are trying to do.

Kevin Pho: So give us a practical scenario from the patient standpoint. When physicians like yourself extend their knowledge using AI, what would that look like for patients?

Tod Stillson: Yeah, so I think at the beginning, you have to have like an onboarding process, right? For me with my own product, we have what we call a symptom checker. That’s an AI-supported, symptom-based intake for urgent care, right? So we’re triaging those symptoms, and our technology is just text-based. We don’t use audio visuals. We use sort of a dynamic text-based process, and it’s just like your nurse does in the clinic, right? When the nurse goes to room the patient, they’re asking the patient, why are you here? What’s going on? They collect that symptom. They collect that history. And what a nurse does for you, medical assistant, whoever your team member is that does this, right? They assemble all of that information. They queue it up for you as the doctor so that when you walk into the room, you have that in your hand. And you go in kind of knowing what the conversation’s going to be about.

OK, now really rural family medicine, because you do everything, the conversation can always veer in a lot of directions, but mostly you’re walking in with that information teed up. What we use is the best of technology in a dynamic way to sort of sort that. And in our system, we sort it into four buckets. Hey, what is a low acuity condition we can treat, number one. Number two, what is this that really needs a face-to-face visit with somebody? It’s not a good history-based sort of decision tree. Third, an urgent or emergent condition that needs to be seen immediately by somebody. And fourth is something viral that just needs supportive care, good evidence-based information, and maybe a work or school note.

So that’s our sophisticated triage system, that with 5,000 patients has been a hundred percent accurate. We’re very proud of that, because we know the recent articles just came out this week about GPT and others that their triage systems are about 50 to 60 percent accurate when it comes to just de novo GPT. So again, comes back to physicians orchestrating and building it.

But for the patient, here’s what it feels like. I get on my phone, I get on the system. I put my symptoms in, and within a couple of minutes I have my answer to those four questions. And in family medicine, because I’ve done this for so long, I used to get asked those questions on the baseball field. At the Walmart, at the church, where they would say these two questions, Kevin, hey, do you think I need to see a doctor, number one. And number two, do you think I need an antibiotic? In my system, I took those two practical questions and organized a simple free triage system to answer those for patients. And it sort of sorts them.

Kevin Pho: And patients are satisfied with that?

Tod Stillson: They’re very happy, right? Because at the end of the day, if they have something that I can easily treat asynchronously in our system, then I can offer that to them. No insurance, cash, price, 25 bucks. So from a patient standpoint, they get care delivered right in their pocket, right where they are, 24/7. They get those answers to those questions, and then they actually have the opportunity to kind of get that simple thing treated if they need to in a simple way. Again, 24/7. That’s patient-centeredness. So that’s what patients are looking for in the marketplace.

To the answer to this question, it’s like there’s a lot of information showing that patients are going to these AI sites and putting in their information, putting in their symptoms, looking for answers, right? And so why don’t we organize and orchestrate something that is high grade, medical grade? By the way, my software is FDA approved, all the yada, yada, yada that go with that. But why don’t we do it the right way to help them? That way they can find an entry that’s simple and patient-centered.

Kevin Pho: Makes sense. So I’m familiar with the study that you cited. It was from Nature. It showed that around 50 percent of the triage information on GPT was inaccurate. Are you concerned that eventually that number is going to go up because, you know, the evolution of these models are exponentially improving? So when we talk this time next year, isn’t it within the realm of possibility that these frontier models are going to be almost as good as what physicians can offer?

Tod Stillson: That’s a fair question. And I think they are going to evolve. They will improve and they will get better. And here’s the reality. There’s not enough physicians to go around. And if that kind of technology can provide sort of a simple solution, not for treating patients, that’s always going to be the doctor-physician relationship, in my opinion. That’s our space. That is not for AI to do. But when it comes to triage, if it gets better and more sophisticated, empower the patients, give them that evidence that they need in the real space, and they can come in and talk to us with that information.

And it’s just like the old days, when I call it the old days, when they’d come in with Google searches, right? They come in, hey, Doc, what do you think about this? I Google searched this. And we go, well, kind of right, kind of not right, let’s talk about this, right? So yeah, it is going to evolve and it’s going to grow. And by the way, our own technology’s going to evolve and grow, because again, you have to have a feedback system when you build AI technology, that it is continually improving on itself. So ours will get dialed in, just like the others will get dialed in. And at the end of the day, I think consumers are the winners.

And I don’t think a doctor should be concerned about that as much, because their need is always going to be there. The human in the loop, the human-to-human interaction, there’s always going to be a need for that, and we can’t be replacing that. But this does hit on the area of provider shortages and physician shortages and PA shortages, and really what we’re used to doing and solving these problems, Kevin, are this. It’s a headcount answer. Let’s scale to meet the needs of the patients by just putting more people in the equation. People are finite. They’re expensive. Frankly, they’re geographically fixed. In the rural areas where I live, you can’t get them to come live here. They don’t want to go to rural areas anymore. OK, so that model of scaling the headcount for a national problem, it doesn’t work anymore. So we need technology.

Kevin Pho: In your article, you talk about one of the potential obstacles to extending with AI, and that’s governance. So tell us what that is and how that could potentially be an obstacle to your vision.

Tod Stillson: Well, governance has to do with the fact that the physicians are involved in the knowledge base. It’s called the KB in the software world, right? So what is that AI or that technology using to source its information to make the decisions that it spits out, right? And for these large LLMs, a lot of these, the sources are public. Some of them more sophisticated on the medical side are feeding on a regular basis, right? The medical information. OpenEvidence would be a good example of that. It’s not set up to be a triage system, but it sources doctors with the most up-to-date information, and we’re going to see sort of a fusion and aggregation of these things that’ll come together so that we’re really, hopefully, all operating out of the best of the best guideline-based medicine.

It doesn’t matter if you’re in rural Indiana or if you’re in downtown Manhattan. You are providing the best level of care using those sources of technology to support your medical decision making in the moment at that right time for that patient. And that’s coming. If we think that medicine is not changing before our very eyes, and it is going to continue to change, and we don’t lean into it, we’re going to miss the boat.

So physician governance is, we’re in the middle of that process. We’re actually helping to create the KB that leads to how these systems work, not so much letting somebody in San Francisco or in Boston who’s really interested in technology, making money and running a business, determining that. We are the ones that are determining that. And that’s that governance. It requires quality and involvements, evaluation, feedback, continuous updating, right? Guidelines change. Our profession changes. And these things ought to be very dynamic and integrated into the ideal use of that technology. So that’s governance, right? It’s not somebody else, it’s us.

Kevin Pho: Now, of course, you’re not the only one that is trying to introduce AI-related health care solutions. There are so many companies trying to capitalize on this space. Not all of them are physicians. Not all of them have a medical background. Do you feel that some companies and people are too quick to frame AI as a solution without respecting what human care actually does?

Tod Stillson: Yeah, 1000 percent. Kevin, you’ve seen the information on MEDVi out there. They’ve been reported, right? So this is a consumer direct-to-consumer company that has kind of done some things, I’m just going to say, unscrupulously when it comes to marketing in the way that they built their system to sort of create a business model. And I think for patients and physicians, that doesn’t do a service. Now, there’s not everything about their system that’s wrong, OK? But when you start cutting corners because you’re trying to earn more money or be a competitor to a market space, that’s where this thing goes wrong.

Kevin Pho: Yeah. So now, for those who aren’t familiar with that story, I know that they were profiled in the New York Times. Just give us a 20-second snapshot for those who aren’t familiar with what they’re doing.

Tod Stillson: Well, at the end of the day, they’re on the GLP-1 bandwagon, so to speak, among other direct-to-consumer products that they’re selling. And in essence, in a couple years’ time, by just putting technology on steroids, they’ve orchestrated a company that stood up almost like a popup in a couple years, and they started doing millions and millions of direct-to-consumer care to consumers that don’t really have the guardrails or safety mechanisms in it that we would expect with physician governance. Now, I’m pretty sure it wasn’t created by a physician. I think it was a non-physician that created it. He saw the market space. He saw the opportunity and dove in. But at the end of the day, the lack of physician governance, the lack of guidelines and ethos involved in the creation of that product, have led to some pretty unscrupulous things that are being identified by regulators. And it’s going to cost them.

Kevin Pho: Yeah. I think I was reading in the New York Times, a billion dollars in revenue just with a handful of employees, if that.

Tod Stillson: Yeah, like, I think there’s 13 employees and a billion dollars in revenue. That’s a revved-up technology. But guess what, here’s what that reflects, though. And I want to point this out to our listeners. That reflects what the marketplace, e.g. the consumers, e.g. the patients, are wanting, OK? That just doesn’t happen with Novo. That isn’t just tricks and mirrors. That is evidence that the demand from the consumer side is so high for cash-base, non-insurance, non-system regulated care. That is huge. And this company was wise enough to tap into it. Yeah, I’m kind of tapping into that as well. That’s why I don’t take insurance with my product. And it’s what consumers want, it’s what they’re interested in. And so when you build a business model, right, you do that, right? There’s a lot of listeners out there that are in direct primary care, concierge medicine, cash-based practices. There’s a whole world out there that exists with this, and consumers are looking for solutions as well as physicians in this marketplace.

Kevin Pho: Yeah. And it’s just an example. If physicians like yourself aren’t tapping into that AI health care intersection, you’re going to have people who are less scrupulous tapping into it, because, like you said, the demand’s there.

Tod Stillson: Yeah, and unfortunately it will happen. It is happening. So let’s have our tribe, let’s have our people get in the game, help bring the governance and bring the right ethos to this, and really kind of create the right system that sort of helps everybody, especially the patients. Honestly, I’m a patient-centered guy. I’ve always been that way, and building and creating something that’s going to help the patient is really my end game.

And you know, that comes back to the origin, the beginning of this conversation, that article, right? The future care teams that we’re going to develop to extend our knowledge base, to take the physician shortage and allow us to grow and provide care for more people. It is not going to be an answer of just add more headcounts. It’s just not there. And especially in the rural areas. I know this for a fact, and it is because we are going to need to use technology and AI to support us in that process, Kevin. I mean, it’s just a fact, and we need to get on board with it, right? These stats, 124,000 physician shortage in 2027, 187,000 in 2037. That affects the rural areas almost more than anything. Urgent care takes a couple hours, and you go sit in urgent care right now, and in the ER, gosh, you could be looking at five to six plus hours, right? Getting providers to come to rural areas is very difficult. The economic reality is labor’s more expensive and the least scalable component of all the care delivery that we do. We’ve got to lean into technology. The moment is now. And I would say to the listeners out there and the physicians, get on board and let’s make this happen for the benefit of our profession and for the benefit of the patients.

Kevin Pho: We’re talking to Tod Stillson, family physician and health care entrepreneur. Today’s KevinMD article is “How artificial intelligence scales physician extension.” Tod, as always, let’s end with take-home messages that you want to leave with the KevinMD audience.

Tod Stillson: Yeah, I just want them to not fear AI. To embrace it and govern it and use it as a capacity multiplier in their clinical practice. Think about the ways that are as simple as the ambient AI taking notes from them, all the way up to building software that allows their knowledge base to be scaled, to reach as many patients as possible with the care and knowledge that they’ve been trained to share.

Kevin Pho: Tod, as always, thank you so much for sharing your perspective and insight. Thanks again for coming back on the show.

Tod Stillson: Thanks for having me on the show, Kevin, and I appreciate it.