WEBVTT 00:00:00.000 --> 00:00:00.917 Chris, welcome. 00:00:01.292 --> 00:00:05.046 Tell me, if you would, how is AI impacting patient engagement? 00:00:05.964 --> 00:00:07.757 It's kind of an exciting time, right? 00:00:07.757 --> 00:00:10.176 Because AI is helping the patient both when they're on 00:00:10.176 --> 00:00:12.512 their own but also when they're with their provider. 00:00:13.013 --> 00:00:16.182 Let's say I'm overdue for a colorectal cancer screening. 00:00:16.224 --> 00:00:19.269 If there's an AI algorithm running on the EMR in the 00:00:19.269 --> 00:00:22.689 background, the doctor not only knows the patient needs the 00:00:22.689 --> 00:00:26.276 screening but knows, oh, I see that you've done screenings at 00:00:26.276 --> 00:00:26.943 home before, 00:00:26.943 --> 00:00:29.529 I see that Cologuard is covered by your insurance, 00:00:29.821 --> 00:00:33.033 therefore, let's do a home screening test instead of, say, 00:00:33.033 --> 00:00:36.286 schedule a colonoscopy—and all that can be powered by AI to 00:00:36.286 --> 00:00:37.746 make the experience better. 00:00:38.329 --> 00:00:42.792 Where do you see the greatest opportunities for AI in this 00:00:42.792 --> 00:00:43.209 space 00:00:43.209 --> 00:00:45.503 but also what are the challenges in reaching that? 00:00:46.629 --> 00:00:48.631 I mean, right, those are two different things. 00:00:48.965 --> 00:00:50.258 Let's start with the second one first. 00:00:50.550 --> 00:00:54.429 So the challenge in health care—what's the number one rule? 00:00:54.804 --> 00:00:56.056 Do no harm, right? 00:00:56.056 --> 00:00:57.474 So do not harm the patient. 00:00:58.308 --> 00:00:59.809 AI is an exciting thing. 00:00:59.893 --> 00:01:03.313 Companies like pharma want to do more with AI to help patients 00:01:03.313 --> 00:01:05.482 and providers, but they're very scared. 00:01:05.482 --> 00:01:08.234 What if it were to go and do something that I wasn't 00:01:08.234 --> 00:01:08.735 expecting? 00:01:09.027 --> 00:01:11.696 You know, typically, they have an MLR—medical, legal, 00:01:11.696 --> 00:01:13.990 regulatory—team approving things very tightly. 00:01:14.240 --> 00:01:16.117 But AI isn't built to do that. 00:01:17.077 --> 00:01:18.286 So that's important. 00:01:18.453 --> 00:01:21.456 And the way to solve it is to have what I'd call medical-grade 00:01:21.456 --> 00:01:21.581 AI. 00:01:21.790 --> 00:01:25.585 So this is LLMs designed with guardrails specific to health 00:01:25.585 --> 00:01:29.005 care, specific to pharma, specific to medical, legal, 00:01:29.005 --> 00:01:29.714 regulatory. 00:01:29.964 --> 00:01:34.052 It does things like make sure that the AI doesn't speak off 00:01:34.052 --> 00:01:38.056 label, it doesn't speak off guideline, it doesn't provide 00:01:38.056 --> 00:01:39.849 treatment recommendations. 00:01:39.849 --> 00:01:42.811 It captures AEs and adverse events in the background 00:01:42.811 --> 00:01:46.147 automatically, sends them off to where they need to occur. 00:01:46.648 --> 00:01:50.193 That's how you build in kind of the safety guardrails to allow 00:01:50.193 --> 00:01:52.612 people like pharma to be innovative in AI. 00:01:53.404 --> 00:01:56.991 What other trends do you see shaping this year ahead? 00:01:57.992 --> 00:02:00.954 So I think one of the things—it's, it's not 00:02:00.954 --> 00:02:05.375 necessarily going to change the world in 2025, but it's starting 00:02:05.375 --> 00:02:08.670 to happen pretty quickly—is AI usage in the EMR. 00:02:09.546 --> 00:02:13.341 One of the leading use cases we're finding that providers are 00:02:13.341 --> 00:02:17.178 loving is that AI can capture the dialogue between the patient 00:02:17.178 --> 00:02:20.140 and the provider and automatically put that into 00:02:20.140 --> 00:02:21.182 structured notes. 00:02:21.182 --> 00:02:24.018 Instead of the doctor having to sit there and type and not look 00:02:24.018 --> 00:02:26.813 at you, the patient, the doctor can now actually pay attention, 00:02:26.813 --> 00:02:29.274 have a dialogue, and let the computer capture all those notes. 00:02:29.858 --> 00:02:32.026 And honestly, it does a better job than they could at 00:02:32.026 --> 00:02:34.112 structuring them in a way that we can then use them, 00:02:34.112 --> 00:02:35.363 we can bill appropriately. 00:02:35.572 --> 00:02:39.534 So that's one big use case. But I think the next big wave is 00:02:39.534 --> 00:02:43.079 having all the clinical algorithms become smart behind 00:02:43.079 --> 00:02:43.580 the EMR. 00:02:43.788 --> 00:02:46.916 So if you think, maybe a simple example, hey, instead of just 00:02:46.916 --> 00:02:50.128 saying, well, you're overdue for a shingles shot because you're 00:02:50.128 --> 00:02:53.173 50 years old and you haven't had one, what if it says you're 00:02:53.173 --> 00:02:56.301 overdue for a shingles shot, but I see you've never taken any 00:02:56.301 --> 00:02:57.802 vaccination that I prescribed. 00:02:58.011 --> 00:03:00.221 I'm not going to bother the doctor with that because we 00:03:00.221 --> 00:03:01.389 don't want more alert fatigue. 00:03:01.389 --> 00:03:03.183 If the doctor wants to address it, they can. 00:03:03.391 --> 00:03:05.518 We're going to instead focus on something else, 00:03:05.518 --> 00:03:07.562 but that's one algorithm that's easy to do. 00:03:07.562 --> 00:03:10.648 What if you have thousands of algorithms running across all of 00:03:10.648 --> 00:03:11.316 patient care? 00:03:11.524 --> 00:03:15.361 Now you need a systemic system, and only AI can have all those 00:03:15.361 --> 00:03:16.696 algorithms be dynamic. 00:03:17.071 --> 00:03:20.867 And the last piece that I would say is, we can go beyond just 00:03:20.867 --> 00:03:23.036 clinical and talk about economics. 00:03:23.203 --> 00:03:26.539 So how do we pull in value-based care contracts into that AI 00:03:26.539 --> 00:03:31.169 algorithm so that when, say, they're doing really well on 00:03:31.169 --> 00:03:34.297 stars, on vaccinations, but not on screenings, 00:03:34.505 --> 00:03:37.759 it intelligently knows how to push forward and put more 00:03:37.759 --> 00:03:41.095 information at the tips of the fingers of the doctor that 00:03:41.095 --> 00:03:42.263 drives the contract. 00:03:42.722 --> 00:03:44.182 Chris, thank you so much for your time. 00:03:44.474 --> 00:03:44.933 Thank you. 00:03:44.933 --> 00:03:45.350 Thank you.