Subscribe
Share
Share
Embed
Vic and Marcus are joined by Dr. Eric Stecker, cardiologist, professor of medicine at Oregon Health and Science University, and co-founder and Chief Medical Officer of Insight Health. They explore the origin story of Insight Health, how generative AI inspired its creation, and what sets their AI apart from typical ambient note-taking tools. The episode covers AI-driven patient intake, EHR integration, improving efficiency in clinical care, reducing physician burnout, and how AI assistants can...
Vic and Marcus are joined by Dr. Eric Stecker, cardiologist, professor of medicine at Oregon Health and Science University, and co-founder and Chief Medical Officer of Insight Health. They explore the origin story of Insight Health, how generative AI inspired its creation, and what sets their AI apart from typical ambient note-taking tools. The episode covers AI-driven patient intake, EHR integration, improving efficiency in clinical care, reducing physician burnout, and how AI assistants can gather nuanced medical histories and support post-visit follow-ups. Dr. Stecker explains how the platform is being used in real-world clinics, its quality assurance safeguards, and what’s next as AI expands into pediatric care and diagnostic support. They also discuss Insight’s business model, physician satisfaction metrics, and the broader vision of transforming population health through scalable clinical automation.
LinkedIn - http://bit.ly/3Hdbq1X
Every week, healthcare VCs and Jumpstart Health Investors co-founders Vic Gatto and Marcus Whitney review and unpack the happenings in US Healthcare, finance, technology and policy. With a firm belief that our healthcare system is doomed without entrepreneurship, they work through the mud to find the jewels, highlight headwinds and tailwinds, and bring on the smartest guests to fill in the gaps.
If you enjoy this content, please take a moment to rate and review it.
Your feedback will greatly impact our ability to reach more people.
Thank you.
Okay.
Welcome everyone to Health Further today.
We have a great guest.
For our guest.
So series, um, is Eric Steckler.
He is the co-founder and chief medical Officer of Insight Health.
Eric, thanks for doing this.
Really appreciate it.
Hey, Vic, thanks.
I really appreciate it.
I listened to, uh, quite a number of your podcasts and they're great.
I really excited about this opportunity.
Excellent.
Yeah.
So, uh, as we start off, uh, maybe give the audience a little bit about your background.
I know you're a cardiologist, but then you have, uh, co-founded a business.
Maybe talk about your practice and then what made you decide to do a, to do a new, new technology startup?
Yeah, thank you.
I'd love to.
Um, I'm a academic cardiologist and actually I've subspecialized in cardiac electrophysiology, um, at an academic medical center.
And you know, over the years I've been in practice for 15 plus years.
Um, I've also been in leadership roles at the American College of Cardiology and seen a lot of the challenges across the country that physicians have faced in.
Managing the huge number of patients who need care and how to deal with the routine aspects of care that really limit their reach, uh, limit their efficiency, limit, their effectiveness.
And so I think every, every clinician, every doctor is, you know, frustrated at times by, by doing work that's routine that they don't otherwise need to be doing.
That could be offloaded to others and could allow them to.
To see more patients and help more people.
So that's the undercurrent, that's what what I, and, you know, many of my colleagues have experienced.
Yeah.
And, and that really was, um, the backdrop of, of sort of how we came to, to this, to this company.
But just to press it a little bit more, I mean, uh, there's a lot of physicians that are frustrated that they are maybe are not practicing at the top of their license or they wish they could hand off more of the.
State entry or, um, record keeping or various tasks.
Sometimes it takes longer to explain what needs to be done than you could just do it yourself.
Um, but talk about like what gave you the impetus?
What was the catalyzing event that you decided, gosh, I need to just go, go try to fix this.
Yeah, absolutely.
Well, yeah, you're, you're absolutely right.
E everybody is frustrated and you know, various people will have different modes of action or impetus to action around that.
And there's nothing particularly unique about me.
I was frustrated as everybody is and would do what I can to improve the situation, but really.
The, the genesis was, um, was on a run, a good friend of mine and, um, uh, uh, one of the other co-founders of the company, Punka Gore, who's a, a neurosurgeon, specialized in brain tumor surgery.
Uh, he and I have been friends for more than 15 years, and we go on a weekly run, and I, I can remember in, uh, December, 2022, uh, where punk and I were running in the dreary Portland weather and, uh.
You know, talking about things as we always do, and this was just a couple of weeks after chat, GPT had been released to the public.
Yeah.
And it was the top of our conversation was how transformative this was going to be for medicine.
It was, yeah.
Totally apparent that generative AI was going to be able to do, to really be able to allow clinicians to leverage themselves in ways that were not at all possible previously.
So we'd both been kind of entrepreneurial before we'd had various ideas, and in this case we said we really need to use this.
This is an inflection point in technology as an inflection point in medicine.
And we need to, we need to be part of this solution.
Um, and so things kind of cascaded from there.
Um, I can get into the details now or, or later.
Um, yeah.
Well, so I think, uh, most of the audience will know about, uh, what's very, I mean basically called ambient note taking or ambient listening.
It's pretty common use case for AI to kind of listen in to the clinic, the doctor patient interaction.
And then effectively do the record keeping, do the note taking.
Uh, but Inside Health does that but, but much more.
Or, or it goes further.
So maybe explain where it's, uh, different than ambient, uh, note taking.
Yeah.
And, and what else you've added in or how you've thought about sort of building out that experience.
Yeah, absolutely.
And you know, I think it's, um, it's, it's interesting to note that we, we did not go first to this ambient note taking use case.
Um.
And I think it's because KO and I are both clinicians.
We're seasoned clinicians, you know, we've been in practice for, you know, more than a decade, and we recognize that, that there's use in that, but there's not the, that's not where, that's not how things are gonna, that's not gonna be the transformative aspect of this.
Um, so we actually moved first into what I think a lot of others have, have been very slow to move into, which is the patient facing aspect of ai.
So.
Note taking.
You know, for those who aren't familiar, the a AI scribe function AI note taking basically has a, a AI recording that AI then processes when that records the patient doctor interaction in the clinic and then summarizes that, forms it into a medical note.
The limitation is it can only summarize the actual conversations that have taken place between the clinician and the patient.
And there certainly is time savings, and that's certainly an annoyance, you know, having to create the documentation from the visit.
Um, and that certainly that is an area for inefficiency, for efficiency.
Yeah.
But a lot of docs have macros or they have little shortcuts that they're already using before ai.
That's right.
It made that less bad.
Yeah.
Yeah, absolutely.
And, and because of that, you'll find there's been varying uptake of AI scribes.
For some people it is absolutely wonderful and essential.
And for others, they have a lot of shortcuts in the nature of their practice.
The nature of their shortcuts has already made them pretty efficient in documentation.
So we recognize there can be value there, but there's not the maximum value there.
Uh, ponage and I recognize that the maximum value is going to be.
Having AI interact with patients, um, in a way that AI can safely and in a way that can offload some of the routine aspects of those patient, interact, patient, doctor, patient clinician interactions, allowing the clinician to then engage in a more meaningful fashion in the, the higher order things that only the clinician can, can, can engage in.
So this does not take the doctor out of the equation at all.
It allows for the doctor though to have a more impactful and meaningful interaction with the patient and to kind of, and, and to be able to, uh, interact with more patients to provide more care, better care to larger populations.
So, um, when I think about, uh, my interactions with the healthcare system, there already are extenders that help the physician.
So, so you, you fill out the form, the intake form in the waiting room.
AI could certainly be involved in collecting that data so I don't have to do the damn clipboard anymore.
Then most doctors have a nurse, um, kind of start off the clinic visit, taking information and gathering data.
Um, where do, where do you, where does Inside Health bring the AI in?
Does it.
Augment those other extenders or how, how does it fit into the practice overall?
Maybe?
Yeah.
Yeah, you're absolutely right.
So just to, to be more specific about, you're absolutely right.
The, the points that you bring up, uh, AI can take over.
So AI can take over from the very front desk.
We, our products can take over from the scheduling of the appointment to the intake.
Paperwork, you know, including the very basic stuff, the HIPAA forms, the demographics, the insurance stuff, um, to the kind of routine, um, clinical information that a form might otherwise capture, but is very limited because it's unidirectional and it's not adaptive in any fashion.
To the more adaptive and clinically nuanced information that a nurse might, might gather from the patient in the, in the exam room, or maybe even a PA or an NP might gather.
And so it can take all of that, it can take over all of those functions.
Um, really allowing those staff to be utilized elsewhere for higher value things and allowing the, the clinician to then have a summary of everything that, that was, that the ai, uh, that the AI gathered.
Um, and then just.
Talk you through how it happens.
The, the clinician then can walk in and look at a summary.
It's kind of similar in academics.
You know, if, if you have a medical student or a resident working with you in academic medicine, that medical student or a resident would go and talk with the patient beforehand, gather all the information, and then come out and present that information to the attending doctor.
Very similar to that.
The, the, the doctor can take a look at the summary, uh, meet the patient.
Thank them for having interacted with the AI virtual assistant.
Uh, verify essential elements of the history that were obtained.
You know, two or three or four things usually are the critical elements to, to verify and then talk with the patient.
Maybe ask a couple, a couple of additional questions, if there's anything additional that needs to be gathered.
And then really start, um.
Really delving into the what, why the patient is there, really connecting with the patient, get, um, talking about what the potential diagnoses are or what kind of testing is required and counseling about those things.
And that's really why we went into medicine and to be able to get to that point rapidly, um, to be able to then extend that to more patients, um, than you could otherwise is, is really a game changer.
Yeah.
And so are you using it in your practice there?
Um, yeah, so we, um, so we have, so the two co-founders, ponage and I, um, uh, ponage is using it extensively in his, in his clinic.
We have for organizational reasons.
Um, I'm not using it in my clinic.
I, um, use it in other contexts.
Punic has uses it extensively in his clinic and as does his, the rest of his partners.
Um, and really we've used that to help to iteratively improve it, of course, but it's really, really improved the efficiencies and, and the patients have, have had great experience with it.
Yeah, that's what I was gonna ask about it.
So how, um.
Like how many has he gone to?
He can see more patients or more per day, or he gets to spend more time with him.
What, what are the, um, metrics that you track?
As far as the impact?
Yeah.
All of the above.
So, and I think that's another, there's no one size fits all and you know, a big part, we haven't gotten to physician burnout.
Clinician burnout, um.
Different organizations will have different priorities for why and how they wanna, um, engage ai AI products.
Um, in, in, in his case, um, he and his partners can both see more patients and get done on time.
And so I, you know, I wish I could share.
Texts.
He, he's gotten texts from partners as well as others, uh, at different, at uh, at different health systems.
People who personally know Bunkage, who've texted him to say it's 5:10 PM and all of my notes are done.
I've seen all my patients.
I'm going home to have dinner with the family.
You know, and, and this is a situation where previously people would be going home to have dinner and still have two hours of charting to do.
Right.
So it really can allow, it can allow for that mix.
So, you know, I don't want people to think of this as.
Another tool, another, another whip to whip the clinicians to have them, you know, continue to work 14 hour days and accept, you know, double the number of patients they're going to see.
The, the use, the benefits of the technology are dependent, uh, or, and can be matched to what the organization's goals are.
And that can include purely.
Clinician satisfaction and reducing burnout because that is a big issue and that will affect healthcare going forward.
Um, it can affect purely productivity or more, more often, a mixture of both.
And then, uh, how do you think about the, what training are you using or are you using one of the existing models and building on top of it?
And then.
Uh, sort of included in that, how do you think about sort of preventing hallucinations or, or bad experiences?
Yeah, absolutely.
That's an essential, uh, question.
We have an architecture that allows us to use any foundational LLM and we, we constantly are evaluating, which are serving the purposes.
And we use different LLMs for different functions are different clinical agents.
And we're constantly monitoring that performance.
It's essential that this be a, a trustworthy endeavor that clinicians and patients can trust is going to be accurate.
And in doing so, we, we have a proprietary technology that evaluates prior to any, any summarization that takes place.
There's an oversight of each individual summary that's created.
We have frequent, um, many times per day, uh, evaluations of every single summary that's been evaluated or that that's been created at another level.
And then we have intermittent spot checks as well.
So we have a lot of, uh, we have a lot of technology.
So do you have like a, an agent sort of.
Internal review, uh, reviewing all of the, uh, maybe everything that was done this morning, maybe in order to check Yeah.
More, more frequent than that.
Yeah.
Yeah, yeah, yeah.
Every, every summary has supervis, supervisory layer and hourly.
We have hourly checks upon those summaries, and then we have spot checks upon that.
So we have a, a hierarchy of quality assurance, and this really ensures that, you know, by the time you get to spot checks, you're, you, you see nothing.
And in fact, mm-hmm.
It really, there's, it's very, very rare that we have problems when there are questions of, I dunno if you wanna call it hallucination.
There's often questions of interpretation.
It's very subtle, very subtle issues, and nothing we've not seen any blatant problems we've seen.
Yeah, I mean, just even with a, with a human, uh, say a nurse, they may have an interpretation of some.
Um, data gathering points that make them collect something else, uh, where you might say that wasn't really needed.
I mean, there, there are judgment calls that aren't, aren't hallucinations, they're just sort of, you know, maybe it wasn't exactly what the doctor would've suggested, but, but it is plausible.
Is that, uh, is that close to how like the interaction could go where the, the, the LOM might be?
Doing something that's plausible, but maybe it's not cutting edge medicine or maybe not what, what you or your partner would choose to do.
Yeah, that's right.
And you know, the, it's, our technology is functioning at a very high level, but you're absolutely right.
It's not, it's not going to be able to gather the clinical history.
For at the level that an expert infectious disease doctor of 30 years would gather, it will, it will be somewhat lower performance than that, but it is very high level.
And, um, in general when, when a doctor needs to ask additional questions, it's the kind of questions that you would expect an expert specialist to have to ask on top of almost anyone else who would've gathered the medical history prior to that.
I also wanna point out that, you know, it's not just, we started out with, uh, patients with, you know, with defined, uh, conditions or, or coming in for specific complaints.
And we've now our technology, we've evolved our technology and now be able to handle patients coming in with no unknown disorders, um, you know, in, into primary care or into urgent care.
And really our, our clinical agents can.
Do a remarkable job of gathering a very effective history without prior knowledge of why the patient's coming in and can incorporate aspects of their medical history to figure out this,'cause this is what a clinician would do.
A clinician would know what the other, what the patient's existing medical problems are, how that might inter, how that might, um, influence what's going on, what medications the patient's on, how that might influence what's going on, what their habits and lifestyle are.
Um, our technology can incorporate all of that to generate questions and, and, and create, um.
Uh, create for the clinician ideas about what the diagnoses may be, though it doesn't actually di diagnose the patient.
Of course.
Yeah.
Yeah.
Okay.
So if, uh, if I'm coming into a primary care doc, it, well, I'm asking, maybe it's the question, does it have access to my medical records in the health system already so that it could gather my past history and when medications I'm on?
And then we both arrive at the, at the clinic visit.
Um, and I have new, new concerns, new symptoms, new things to talk about, but it has my history already.
Is that correct?
Absolutely, yes, absolutely.
With integration with EHR integration, it will have access to that and it, that will be used to inform what questions are asked based on what may be going on, because it's very different.
A kind of a blank slate with, with regard the same symptoms in the context of a blank slate, an unknown medical history.
Um.
Will be very different and can be from different reasons.
And, um, there therefore different questions would be asked by any clinician, AI or human than if you know actually what's going on with the patient.
So yes, it does have access to, to your medical record through the EHR and the clinician does as well.
And our technology can provide summaries for the clinician.
Uh, from prior notes to, um, to help the clinician when they walk into the room understand, um, more about the patient's background without having to go through 60, 80, 90 pages of background information.
Right.
Okay.
Um, so it sounds like the clinicians, the docs are really liking it.
How are the patients found it?
Have you got feedback from the patients?
Yeah, we, we have a lot of feedback, a lot of positive feedback.
I should say that we have, um, we have research, we have a, um, uh, a research protocol approved, uh, by the institutional Review board, IRB of a top 10 university.
And that will hopefully be launching soon, uh, looking at the feasibility and, and comparing our technology to, um, to humans gathering, uh, to human doctors, gathering information from patients.
Um, and then, uh, uh, we also are, are in discussions with some health services researchers about utilizing our extensive.
Uh, database of anonymized data to, to glean how the patients more formally glean and objectively glean how the patients have, have, have, have experienced it.
We have some just delightful stories.
Um, I can remember one patient with, you know, chronic pain.
Um, you know, if you have chronic pain, if you've suffered from chronic pain for a lifetime.
Sometimes it's therapeutic and important just to be heard, just to know that somebody is listening to you.
And unfortunately, we're also busy.
Clinicians are so busy that we don't often have that time.
Well, this patient engaged with our, our clinical assistant for over an hour talking about their pain history.
Yeah.
Talking about their symptoms and Yeah.
And, and it was very patient and, and went through as much detail as they wanted to go to.
Exactly.
It's really, it's really gratifying to see, and the patient, you could just tell from the transcript that the patient came away from this, you know, just very pleased.
And, and, um, another, another instance I can think of, uh, just from some of the quality assurance reviews I've done is a patient who clearly, we don't personify this with, our clinical agent is named Lumi.
So to that degree we have, but we've not, otherwise we, there aren't, there aren't, um.
There aren't emoticon or there aren't, um, pictures.
There's not, it's not really personified beyond having the name Lumi, but this patient, um, you know, this patient was interacting very con, very comfortably with the AI agent and.
When they, when they would ask questions, um, uh, he would respond, yes, ma'am.
No, ma'am.
So clearly very comfortable and a, a delightful experience for that patient as well.
Yeah.
Okay.
Excellent.
Um, now I, I'm, you know, I'm a podcaster at my day job as a venture capitalist.
I can't help but ask, like, have you, uh, brought the pro product out to market?
Have you raised capital?
What, what's the, um.
Business model, how do you charge for it?
What's it like?
Yeah, yeah.
We sure have.
Um, so we, I, I'm really, really proud to say that we were the first autonomous patient facing, uh, AI agent, um, to, to be, to be released and so late.
So Jet Chat, GPT came out end of November, 2022.
Our first patient interaction, autonomous patient interaction was at the end of 2023.
So just 12 months later.
Yeah.
Um, we since then have licensed.
To multiple clinics, um, mid-market clinics, uh, large sub large specialty clinics, and we we're now deployed across specialties, across primary care.
We have, um, our technology is used to gather intake information for patients before visits.
It's used to summarize the information during visits, both for the clinician's medical note and for for patient, um, for patient, uh, after visit summaries.
And it's also used to check in with patients after the visit.
And, you know, I think that's really an under-recognized issue is we kind of think about healthcare a as it is now.
Um, and we kind of think that's the best we can do.
But if, if you think about it, ev every patient, so on, you know, let's just do something simple like high blood pressure.
So you're an otherwise healthy person who goes to the doctor and you're diagnosed with high blood pressure.
You started on a medication for high blood pressure.
In an ideal world, you would get a call, you know, or you would be contacted a week later and just checking in.
Say, Hey, did you fill the prescription?
Did you start the medication?
Were there, were there side effects?
Did you know?
How are you taking it?
You know, what?
What are your experience like?
Are you taking the medication or did you get the labs that were supposed to be drawn after you started it?
Uh, we really should be doing that for every patient, right?
But we do that for almost nobody can afford to do that.
We do that for pa very complex patients who are hospitalized six, 12 times a year.
We can do that for, because the costs for them are so high.
But for the average patient, just with a new diagnosis of hypertension, we can't afford to do that currently with technology like we have at Inside Health.
Um, you can do that.
Every clinic could do that for every patient starting a new medication.
Yeah, it's interesting.
Uh, I'm, I'm in the midst of writing a book right now.
Uh, and the, the, the book ends with, uh, kind of a goal to try to end heart disease, which will be close to you.
And I think part of the way to end heart disease is to provide everyone who has the early signs of heart disease in their thirties with an AI personal health coach and, and, you know, much more careful.
Cardiac care, uh, because we have a lot of tools to, to manage cardiac disease.
If A, we catch it early and b, we sort of bring it to the people so that they engage more.
Um, it's really exciting to see like what could happen if we gave every person this per this one-on-one attention.
You are absolutely right.
And, and you're thinking of it from the consumer perspective, um, which is great.
And, and for engaged.
Um, patients or people, um, that can do wonders, but a lot of people aren't that motivated, aren't that engaged, and they won't do that.
And for them, we need to rely upon healthcare professionals to shepherd them along and it ends up right.
This was actually when we, when we, when we It's probably all of the above both end.
Right?
Like we need it, it really is important to need to power the, the, the clinicians.
Yeah.
It, it's essential to have both.
And that was actually our initial conversation.
I think when Punga and I, as co-founders took this idea to our other two co-founders, um, Jay Sson, Andron Siva.
That's what we, that's the vision that we created for them, is that we can use AI to extend what clinicians do in a way.
To do the routine things that we know need to be done that can save millions of lives a year from heart disease alone, for instance, by simply reengaging with a patient about their smoking.
Are they, are they ready to consider quitting yet?
Mm-hmm.
No, they aren't.
Okay.
Well, we'll check in with you again in six weeks.
Yeah.
Ha.
Again, the hypertension.
Have you started your medication?
Are you taking it?
Are you having side effects?
These simple things.
Medication adherence, smoking, taking aspirin, if you're supposed to be taking aspirin, getting your cholesterol checked.
These simple things, we don't have the bandwidth in healthcare right now to do.
We just see the patient in the office.
Yeah.
And the, and the pay for a human to jump in the, the payoff it doesn't receive enough money for, for you or even one of your nurses to fine me and engage with me.
But if you can do it with a technology that's.
Pennies it, it could really be good.
That's exactly right.
And it can, and it can save literally millions of lives.
So that, that's the kind of, that's the big vision and that's why we got into it.
And I think that's where we're going to go.
Okay, so you already mentioned that it's agnostic, agnostic to the different frontier models.
So you can, you can pick and choose and if a new one came up next month, you could add that, you could knock one off.
Absolutely.
Um, does it sort of take learnings from the field?
Like you have, I don't know how many instances out there working with physicians and patients and then pull back maybe exceptions or learnings?
And gets smarter over time as it engages with patients and physicians.
Absolutely does.
Yes.
We, we absolutely, um, we improve the model and our interaction with the models based on, um, the experience with the clinicians.
Um, we have a lot of engaged clinicians who give direct feedback, uh, kinda in a formal manner.
We'll click within the app and, um, give us feedback on summaries on what, what should be added on what they didn't like.
Um, we can also do that without active clinician, uh, feedback.
We have partners.
I'm really excited.
We, we have a, um, we've started with, with some really good channel relationships now, including with EHRs, specialty EHRs, and with an EHR partner.
We're gonna do some great work going forward to really refine this for pediatrics and really, really tailor.
How the, how the interactions, how the model performs for, for pediatrics, you know, to address gaps in care, to address some of the subtleties and, and, and special considerations for, for kids.
So we, we do this, uh, on a routine basis across specialties, and we can do so in a focused manner for specific areas.
Yeah.
Okay.
Excellent.
So over time it'll get better and better.
Uh, as far as what it, what it is good at covering and, and maybe understanding different.
Different, um, exceptions or different use cases for this?
Yep, absolutely.
Beyond the foundational model work, um, much more improvement on top of that with what we were doing.
Yeah.
Okay.
And then what's the, how do you charge, like, what is the physician pay?
Is it per engagement, is it a monthly charge?
Like how does it, and how does it fit into the, to the practice of medicine right now?
Yeah, absolutely.
There's a, there's different models for different, um, uh, implementations of it.
Okay.
Um, there are some referral management, um, that are based on, on a number of processed, successfully processed referrals.
For instance, for preventive care, for colonoscopy.
And other things, that's kind of a referral based processing of a queue.
Um, but for kind of the average clinician, it's kind of a per seat, per month basis.
Um, okay.
For the average clinician who's using it for, as I'd mentioned, for intake, for in visit summarization, yeah.
For patient follow up.
Yeah.
Okay.
So, and it's at a price for it, where it is, um, the change in their happiness and job satisfaction is such that, uh, they're, they're much more satisfied and they can see more patients and thereby generate more income for the practice.
Is that close to right?
Yeah.
Um, the ROI conceptions can be different for different clinics depending on the goal.
So for some, if it can improve satisfaction or reduce turnover, if there's been a lot of turnover among PAs, nps, MDs, um, it, it will, that will be the ROI.
Um, for, for others it can re, you can repurpose, um, uh, staff PAs, RNs to do other things that they had and allow the AI to take over what they had been doing.
Um, so it's really gonna be a different ROI, it's gonna be very context specific.
Okay.
Excellent.
Um, okay then you've been early in sort of adopting and implementing AI into your practice and other your partner's practices.
Where do you think the future's going?
Give us a little bit of your view of kind of where this is heading in 12, 24, 36 months.
Yeah, I, I think that, um, I think that our vision of an AI platform to offload routine care with the patient facing aspects and the documentation aspects of routine care across the, across the whole spectrum of care is, is where we'll be in 12 to 24 months.
So from the very, from the first moment the patient registers with a clinic or hospital to the prior authorization that happens for procedures or testing that's done to follow up all aspects of this being centrally handled by ai.
Clinical assistance.
I think equally important is the, the diagnostic ensuring appropriate care.
There are gaps as you know, clinicians, you know, most, most medical.
Most medical conditions aren't a diagnostic dilemma on tv.
They appear to be, but you know Yeah.
It's pretty clear what the, what the.
Diagnosis is, yeah.
It's sort of how do we get the, get the treatment, and how do we ensure we actually do the treatment?
That's exactly right.
The, the thing that is routine, unfortunately, is that there are gaps in care.
So we've made the diagnosis of Corona artery disease, but we're not, we're not treating it according to, to guideline based care.
Mm-hmm.
And so we can, this is also an area that, that, that our technology, um, can really work to improve, and that is to elevate for clinicians, Hey, your patient, you diagnosed your patient with this condition.
We've noticed that they are not on a, a statin or they aren't on a high enough dose of a statin.
Um, so it can very effectively do that.
It can also, our technology can also make diagnoses, but that's, um, that's farther in the future.
That's farther down the roadmap, I think for us and for everyone else.
Because for a variety of reasons, there's, there's more risk there.
Um, she is it, where do you cross the line into a regulated medical device, software as a medical device.
Um, other issues I've not heard people talking about that are very important, like anchoring bias.
Um, you know, the first thing, um, both Punji, other co-founder, chief medical officer, and I, both of our wives are, are doctors as well.
And, uh.
You know, the first thing my wife mentioned about, um, decision support, diagnostic decision support is watch out for anchoring bias.
Watch out for the clinician.
The first thing you show the clinician as a potential diagnosis is what they are going to start thinking about.
And, and, and that may not be what you want.
So how you handle that for, for different conditions and for different clinicians in different contexts is also very important.
Yeah, and it's interesting.
There's sort of a. I mean, there's at least two, maybe three different considerations is like, what can we do from a technical capabilities point of view?
Then there's what is regulat, what's what's allowed in the regulatory environment and what, and what's reimbursed?
And then there's the adoption of like, how are the physicians and patients willing to, to change their behavior?
And those things don't move at the same pace.
I think the technical capabilities are far ahead of.
What is regulated and what is accepted by both the physicians and the patients right now?
You're right, you're right.
That's absolutely right.
Yeah.
So you, so you think within 12 months there'll be health systems that have a kind of an ai, um, platform that is helping the patients maybe navigate through the system and advising both the patient.
And the physicians and the whole care team along the way.
That seems, I do very optimistic, but I would love that.
I do, I do think that's quite possible.
It's technically possible.
Um, is it commercially and operationally possible?
Um, I think that's the bigger wild, yeah.
It's, it's much more the adoption curve than the tech capability that I think is, is optimistic.
No, I think it may be a little bit optimistic, but I have to say that.
You know, our early customers have been mid-market, have been, you know, large multi-specialty clinics, that sort of thing.
Mm-hmm.
They can move, they're somewhere between, they have financial resources, they have the, they have the vision, but they don't have the bureaucratic inertia that enterprise health systems have.
Yeah.
And not as much, and so.
Maybe it's over overly optimistic, but I do think that, uh, that, that customers like that could see these benefits, could have the financial wherewithal to, to, to do it, see the financial benefits from doing it, and have the quick turnaround, um, uh, from an operational.
Yeah.
And then, and then you might get the positive sort of pull where they start delivering much higher quality care and they have a better work experience so they get better, more talent.
So then it kind of, in a positive peer pressure way, it forces both big and small.
Uh, provider groups to, to come on board and and adopt.
We sort of, you're absolutely right.
And you know, I didn't mention from an ROI perspective there, there are quality incentives, um, from.
Medicare and other, and, and private payers.
There are quality incentives that our, our technology can really be used to maximize, you know, so they're really nice.
They're really win, win-win, you know, better quality care.
Yeah, I mean, just some of the close-knit care gaps, there are easy layups to do.
Yeah.
Yeah.
And so if you.
If you tune things, if you tune the implementation, the technology, and the implementation for that, you can generate ROI just by improving care across the spectrum of care.
And I think that that can be, that can be realized, those financial benefits can be realized by, you know, larger multi-specialty groups and primary care groups and, um, and I think that could do it.
Yeah.
Excellent.
Well, Eric, where can the listeners find you Is, uh, is their website, where should they learn more about Insight Health and, uh, get signed up and, and begin to help this, uh, adoption over the next 12 months?
Yeah, absolutely.
Love to see, see them.
Come to our website anytime, insight health.ai and, uh, come in peruse some of the use cases that we currently have, how the technology works.
They can sign up for.
Our freemium AI scribe, although as I, as I've mentioned, that's only one small part of everything that we offer.
And, um, yeah, love to see them.
Uh, love to see them explore it and contact us at any point with any questions or interest.
Yeah.
Great.
Anything we didn't cover that, that I should have, uh, asked you about?
No, I think you, I think you asked me a lot of the most important questions and I think anything else you might have asked me I might not have an answer for.
So yeah.
Okay, Eric.
Well thanks for doing this.
Really appreciate it.
Exciting stuff.
Coming outta the West Coast and we'll have to have you back on to sort of get an update in the next couple months.
Yeah, I'd love to.
I'd love to.
Thanks for your time.
We're really excited to be on.