Semantic Health & the Future of Medical Coding with Dr. Nicola Sahar | [Sidecar Sync Episode 94]

Summary:

In this episode of Sidecar Sync, co-hosts Mallory Mejias sits down with Dr. Nicola Sahar to explore how a University of Toronto med student dove head-first into clinical AI research and went on to found Semantic Health, an AI-powered medical coding platform later acquired by the American Academy of Professional Coders in 2023. They unpack the art and science of medical coding, discuss why AI will augment rather than replace coders, dive into the critical role of data privacy and private-cloud deployments in healthcare AI, and share why associations are uniquely positioned to guide their members through this AI transformation.

Dr. Nick Sahar is a Canadian physician and health tech entrepreneur who co-founded Semantic Health. In 2023, Semantic Health was acquired by the American Academy of Professional Coders - AAPC - marking a significant milestone in the adoption of AI in healthcare operations. Today, Dr. Sahar remains an active voice in the healthcare AI space, focused on building tools that support clinicians and improve patient care.

LinkedIn - https://www.linkedin.com/in/nicolasahar/
Semantic Health - https://www.semantichealth.ai/
AAPC - https://www.aapc.com/

Timestamps:

🎉 Thank you to our sponsor

https://cimatri.com/

📅 Find out more digitalNow 2025 and register now:

https://digitalnow.sidecar.ai/

🤖 Join the AI Mastermind:

https://sidecar.ai/association-ai-mas...

🔎 Check out Sidecar's AI Learning Hub and get your Association AI Professional (AAiP) certification:

https://learn.sidecar.ai/

📕 Download ‘Ascend 2nd Edition: Unlocking the Power of AI for Associations’ for FREE

https://sidecar.ai/ai

🛠 AI Tools and Resources Mentioned in This Episode:

ChatGPT ➡ https://chat.openai.com

Claude ➡ https://claude.ai

👍 Please Like & Subscribe!

https://www.linkedin.com/company/sidecar-global

https://twitter.com/sidecarglobal

https://www.youtube.com/@SidecarSync

Follow Sidecar on LinkedIn

⚙️ Other Resources from Sidecar: 

• Sidecar Blog
• Sidecar Community
• digitalNow Conference
• Upcoming Webinars and Events
• Association AI Mastermind Group

More about Your Hosts:

Amith Nagarajan is the Chairman of Blue Cypress 🔗 https://BlueCypress.io, a family of purpose-driven companies and proud practitioners of Conscious Capitalism. The Blue Cypress companies focus on helping associations, non-profits, and other purpose-driven organizations achieve long-term success. Amith is also an active early-stage investor in B2B SaaS companies. He’s had the good fortune of nearly three decades of success as an entrepreneur and enjoys helping others in their journey.

📣 Follow Amith on LinkedIn:
https://linkedin.com/amithnagarajan

Mallory Mejias is passionate about creating opportunities for association professionals to learn, grow, and better serve their members using artificial intelligence. She enjoys blending creativity and innovation to produce fresh, meaningful content for the association space.

📣 Follow Mallory on Linkedin:
https://linkedin.com/mallorymejias

Read the Transcript

🤖 Please note this transcript was generated using (you guessed it) AI, so please excuse any errors 🤖

[00:00:00] Intro: Welcome to the Sidecar Sync Podcast, your home for all things innovation, artificial intelligence and associations.

[00:00:14] Mallory: Hello everyone, and welcome to today's episode of the Sidecar Sync Podcast. Your home for all things innovation, artificial intelligence. Associations, technology, and probably everything in between. My name is Mallory Mejias and I am one of your co-hosts along with Amith Nagarajan, and today we've got an interview edition of the Sidecar Syn podcast lined up for you.

[00:00:37] We're going to be. Answering the question, what happens when a medical student decides to build AI instead of practicing medicine? That's exactly what today's guest did. Dr. Nicola Sahar. He's a Canadian physician and health tech entrepreneur who co-founded Symantec Health. After witnessing firsthand the heavy administrative burden placed on clinicians, [00:01:00] Nick led the development of AI powered tools to enhance medical coding, auditing, and clinical documentation.

[00:01:06] In 2023, Symantec Health was acquired by the American Academy of Professional Coders or A A PC marking a huge milestone in the adoption of AI and healthcare operations. Today, Dr. Nick Sahar remains an active voice in the healthcare AI space, focused on building tools that support clinicians and improve patient care.

[00:01:28] So what are we gonna talk about in this conversation? Well, of course we're gonna talk about how. Dr. Nick Sahar went from medical student to AI entrepreneur, ultimately selling his company to A A PC. And what this acquisition tells us about association's leading innovation rather than following it, we will hear his thoughts on AI replacement, fears and medical coding, a profession that does seem a bit vulnerable to AI and why he believes it's about augmentation, not automation, and what happens ultimately to professionals who don't adapt to ai.[00:02:00]

[00:02:00] We talk a lot on this podcast about how important data privacy and safety is. I can't really imagine an industry where data privacy is more paramount than healthcare. So we're gonna talk about that with Dr. Sahar and then we're gonna get his perspective on why he believes AI will be one of the most impactful technologies ever in our species history.

[00:02:22] And ultimately why associations are uniquely positioned to guide members through this transformation, everybody. Thank you for tuning into today's episode. Please enjoy this conversation with Dr. Nick Sahar, Nick Sahar. Dr. Nick Zahar, I should say thank you so much for joining us on this Sidecar Sync Podcast.

[00:02:42] As we kick off the episode, I'm hoping that you can share a little bit about your background from medical student to now, where you've been, where you're going, and kind of. Everything in between.

[00:02:53] Dr. Nicola Sahar: Yeah, of course. Thank you so much Mallory, for having me on. Um, as you mentioned, I did start out as a medical [00:03:00] student in Canada at the University of Toronto, and uh, I think from week one of med school, like just jumping in, I got interested in the tech side of things, uh, at UFT.

[00:03:12] They have a huge AI kind of department with, with a lot of, uh, a lot of the grandfathers of AI actually started out there, like Jeff Hinton and so I just got sucked in from week one into the AI space. And uh, uh, I just dove in. So I, I didn't have an AI background before I knew how to code. From, from my undergrad days.

[00:03:38] Uh, but, but really in medical school, like one of the things that I started doing was looking at clinical natural language processing, which is how do we build like machine learning and AI models for clinical text. Uh, so patient text and, um, I started doing research. On those topics. So I joined a lab, uh, [00:04:00] that did AI research back in, I would say like 2016.

[00:04:04] So a long time ago before chat GPT came out and, uh, I, I started looking at how do we build these models so that they can better understand clinical text and better, uh, be able to, uh, pick out important pieces of information from our clinical nodes. Like diagnoses or procedures or things that happened to the patient.

[00:04:26] Uh, so that was happening on the side. I did my medical school training, but really as I was going through med school, I became like a lot more interested in this technology side than, uh, what I was doing in medicine. Um. So at some point I did a program called the Next 36, uh uh, which was, it's, it's designed as a program to bring Canadian students together on campus every year and essentially puts you in, in like a dorm.

[00:04:59] For [00:05:00] one summer, and then your whole goal is to build a startup, uh, is find, you know, two or three team members and just build a startup. And most of, most of the startups don't end up making it past the, the program. But really the goal is to kind of put you in that environment and give you the energy and, and the inspiration to actually like, have the belief that you can build a company.

[00:05:24] Uh, so. I went through that in my second summer of, of medical school and then I went back to, to the hospitals and rotations and I just couldn't like get that experience outta my head, uh, at the end of the day. So very slowly over the next like year and a half, I've built a lot of conviction over starting a company and building my own thing.

[00:05:48] And uh, so I was rotating in the hospitals, like working in the different specialties in the fields. Then really paying attention to problems and challenges that I was seeing [00:06:00] and where technology and specifically the AI that I was researching could be impactful. And, uh, slowly but surely, I, I kind of. Got a lot more clarity and vision about where I would start.

[00:06:15] And it's really, um, really in the back office of healthcare. It wasn't in the front office. There's a few reasons for that we can dive into, but the technology, as I saw it was still early in ai, but there was a lot of manual and kind of error prone. Workflows that were happening in the back office, in the revenue cycle or health information management space that I thought we could apply the tech to now to create, uh, huge impacts to help people on the ground augment how they're doing things.

[00:06:49] So we started with medical coding, which I can get into, uh, a little bit later. But medical coding is essentially the process of, uh, [00:07:00] a hospital. Submitting a claim on behalf of a patient that summarizes everything that's happened to that patient and during the hospital stay. So what are all the procedures that, that the patient went through?

[00:07:13] What are the diagnoses that they've been diagnosed with? Um, we have to accurately capture that so that the, the hospital can get accurately reimbursed for what they did. And the way that happens is usually a team of people called medical coders are reading through all of the clinical notes that a doctor will have typed, uh, for a patient.

[00:07:35] And, uh, based on that, they create this claim. So we started to apply AI to that. Process, uh, to help augment these medical coders. And that's the, that was the initial seed, uh, of inspiration for the company, uh, and to, to make things super short. Um, I didn't end up doing a clinical [00:08:00] residency and I just dove headfirst into this company and then built it.

[00:08:04] From 2019 until acquisition, uh, by A A PC, which is the American Association for Professional Coders, uh, in 2023. So I will stop there. I know it mean, what a

[00:08:17] Mallory: fascinating story. No, it's, it's so crazy, right? How life takes you on journeys that you could have never imagined. You went to medical school thinking.

[00:08:24] You'd be a doctor, of course. And you are a doctor, but you were led on this different pathway through that challenge. I think you mentioned it was called the next 36. Yeah. I have to ask, were the seeds for Semantic Health planted in that competition or, or did the startup you come up with, was it something totally different?

[00:08:42] Dr. Nicola Sahar: I don't think the exact like vision for Seman was planted there, but the, the, the conviction that I could. Build something as a doctor that knew pretty much nothing about business or entrepreneurship, uh, that that was born in that program. So like I was [00:09:00] exposed to a lot of other entrepreneurs, like super successful ones that were Canadian that came and talked to you and very smart, uh, students from across the country.

[00:09:10] And I think it really just shifted the mindset of. Of being in a lot of my, my friends in that program too. We can actually do this and we can actually go build stuff and make an impact and you don't need much, but like conviction and self-belief and, uh, honestly, like a lot of relentlessness.

[00:09:30] Mallory: Indeed, a lot of relentlessness in the startup world.

[00:09:33] Amit, my cohost, is a big fan of hackathons. He's actually going to one next week, and I think it's, it's a similar concept, just being together with other people, feeding on that energy and taking the time away from busy life, professional, personal, and dedicating some time to. To innovating and building things.

[00:09:50] So I think that's really neat. The idea of the next 36, uh, you, you kind of gave us a high level overview of medical coding, which I realized as I was prepping for this [00:10:00] interview. I don't know a ton about, but this, it's always a great opportunity for me to learn. And I'm sure we have listeners that maybe don't know a ton about medical coding as well, so I wanna talk about what that looked like pre ai.

[00:10:10] And you kind of covered this, so you have a doctor provider writing a note after they meet or while they meet with a patient. About like the diagnosis and treatment plan, whatever that may be. And then you have a team of medical coders that traditionally would look at that note and would write the codes, create the codes.

[00:10:30] Yeah. Code to submit. Submit to the To code. Code to submit to the insurance company. Is that correct?

[00:10:36] Dr. Nicola Sahar: That's pretty much it. I would say the only other caveat is before electronic health records, all of this was done on paper. So the team would be pretty much in the hospital, like usually in the basement or Wow.

[00:10:48] The health information management department. And there would be a bunch of paper records everywhere that there would be reading through. But with, uh, the introduction of EHRs. All of that data was [00:11:00] digital, so, so it's a lot easier now for medical coders and also for AI to, to read through this data.

[00:11:07] Mallory: Okay.

[00:11:07] And so Semantic Health comes in at the point of the provider having written this note, and then you have AI look at the note and code correctly based on maybe a set of parameters that you've put into the model or something like that.

[00:11:21] Dr. Nicola Sahar: Yeah, so coding is actually like a bit of a science and a bit of an art.

[00:11:26] Uh, so, so you can't, there's, there's no like, set of rules that you could give through a model that perfectly get every single code because there's a lot of variability in how the documentation that a doctor writes can show up. They can make mistakes, they can write things in a very vague or unspecific way.

[00:11:44] They can miss, uh, including diagnoses. Um, and then the rules always change. There's. A hundred to 150,000 different medical codes that you can assign, uh, that're supposed to cover every single thing that could go wrong and [00:12:00] every single procedure that you could do in a, in a hospital. So. There's so many codes.

[00:12:06] There's variability in the documentation, and then there's a lot of rules for how to apply codes. Like when does a code apply to this patient, when does it not? Um, and it's all dependent on the patient context and what the doctor has written. So it, it's actually a very difficult problem for a person to do this accurately.

[00:12:25] Which is why there's a whole other industry around auditing and clinical documentation improvement, which is basically like, how do we correct the coding and the documentation to make it accurate? Uh, and then it becomes really hard for AI to do this accurately as well. But, uh, when you marry the AI and the person together, uh, they become a lot faster and more accurate.

[00:12:49] Mallory: Hmm. That human in the loop augmentation, that's Yes. Fascinating. And I'm assuming the outcome of this of less errors in medical coding is [00:13:00] beneficial for the hospital and then beneficial for the patient as well. Can you speak to both of those?

[00:13:05] Dr. Nicola Sahar: Yeah. I think the core reason, like ultimately like why we started Symantec is the data itself and healthcare is not necessarily reflecting what happened to the patient.

[00:13:17] Like you have no guarantee that the doctor has written the documentation accurately. Or the codes were assigned accurately. And if you don't have a clear lens into what actually happened to the patient, it becomes hard to action on that data and build other AI on top of it. So my, my initial vision for Symantec was how do we make sure that the data is as accurate and as clean and as structured as possible so that when AI eventually takes off, we have much cleaner data for us to actually build on top of and enable ai.

[00:13:49] So that's like the long-term benefit. Mm-hmm. The short term benefit is the more accurate the, the data is on a patient, the more accurate the reimbursement is for the hospital. [00:14:00] So if you're missing diagnoses, codes, you're underreporting the complexity of care for a patient or the complexity of what you've done.

[00:14:09] So you make essentially, you, you, you, you've already spent that, that amount of money, but you're getting back less. In the same way, if you're over-reporting, that's also not, not beneficial either, because on the payer side, they're paying more than they should have at the end of the day. Mm-hmm. So the goal is accuracy and quality to reduce overpayments and underpayments.

[00:14:33] And then ultimately with cleaner data, the hospital can just make better decisions for patient care. Like the physicians just better understand what happened in the previous discharge. Mm-hmm. What happened? With this patient, population health can, can be better informed with, with cleaner and more accurate data.

[00:14:51] Uh, and then you can just make better operational decisions. So the data really it, the medical coding process actually like [00:15:00] sees all of the clinical data that a patient. We'll have accumulated in this, in the journey of being in the hospital. So it was like the one step in the, the, the whole backend cycle that actually gets all the clinical notes, and that's why we decided to, to really like, intervene there and, and help augment that process.

[00:15:21] Mallory: I wanna dive into the fact that Symantec Health was acquired by A A PC, which you mentioned in what year?

[00:15:27] Dr. Nicola Sahar: 2023.

[00:15:29] Mallory: Okay, so, so fairly recently, not in the world of ai, but kind of relatively, fairly recently, and as you mentioned, Nick, A A PC is the nation's largest education and credentialing organization for medical coders, billers, auditors, practice managers, documentation specialists, compliance officers, and revenue cycle managers.

[00:15:47] I'm curious, how did that partnership come about and what made a A PC the right fit for Symantec Health?

[00:15:56] Dr. Nicola Sahar: So it, it is an interesting story, but we, we started [00:16:00] out as technology partners actually. So A A PC be, besides being the credentialing like powerhouse of the coding industry, uh, actually has, um. Really awesome software for coders.

[00:16:13] So codify, for example, grouping software, educational resources. That's a growing business line that A A PC, um, that A A PC has invested a lot into. And we were looking to integrate the Semantic platform with the grouping software and with the codify software. So we started out as integration partners. We were looking to integrate.

[00:16:35] Partner and, and go to market with a combined kind of technology offering. And that quickly evolved into a lot more because on our side, uh, what was really exciting was a PC was, was really forward thinking. So they were really thinking about, you know, AI is coming, medical coding is going to be one of the first places in healthcare where AI is going [00:17:00] to penetrate.

[00:17:01] How do we actually. Think about the future of medical coding. What does a future generation of medical coding look like? What does a coder look like with ai? Like, these were all questions that they were thinking about 2, 3, 4 years ago before AI really, really took off. And so they were forward thinking, they were trying to answer these questions and, and really be the thought leaders and, and the innovators with how we actually marry AI with, with medical coding, with auditing, with CDI.

[00:17:32] That was exciting. Uh, for them it was because they were trying to answer these questions and, and they were really focused on, um, the technology side of the business. We were an exciting opportunity to accelerate. Their, their, their, their vision in, in the AI space and in coding, auditing, CDI augmentation.

[00:17:56] So we started talking about what it would look like to [00:18:00] partner together and what we would do. And very quickly that, uh, accelerated into. Acquisition conversations and, uh, and really getting to know each other a lot, a lot more deeply and, and understanding, you know, each other's cultures, each other's mutual fits our vision.

[00:18:18] And it all ended up aligning, uh, at the end of the day. So we, we both went for it.

[00:18:23] Mallory: That's super exciting and congrats to you on that. Are the members of A A PC, individuals like individual coders or their organization memberships as well?

[00:18:35] Dr. Nicola Sahar: The majority are individuals. Okay. Uh, some organizations will have like broader memberships for, for individuals, but the majority are going to be individual coders or auditors or, or, um, people working in the, in the revenue cycle that, um, initially get credentialed through a pc.

[00:18:53] So you can go through and credential, take an examination, get certified to actually do medical coding, [00:19:00] and then it's, it's more of a community after that. Mm. You can go to the conferences, uh, complete webinars, credits, educational credits, and um, and meet other people in the community. Uh, so there's local chapters, there's, you know, national conferences, there's a bunch of stuff that you do as part of the organization to, to stay involved at the end of the day.

[00:19:22] Mallory: Mm-hmm. I feel like across the board, what we hear at Sidecar is we have associations where their members are loving AI, already using AI in their workflows. I'm thinking actuaries being an an example of that. Yeah. And then we've got many associations where their members are absolutely put off by artificial intelligence.

[00:19:43] I'm thinking writers, creatives. Right. And so, and then those in between. I'm curious, how did A A PC and Symantec Health see medical professionals? Reception to utilizing artificial intelligence in the coding process?

[00:19:58] Dr. Nicola Sahar: Yeah, I think it's also mixed, [00:20:00] like with any, mm-hmm. Like there's a lot of medical coders out there and, uh, various different, uh, experiences and people at different parts of their career.

[00:20:09] I think, I think there's a lot of excitement and a lot of curiosity about where it can go and how it can help. And there is definitely a portion, a large portion of people that. Are open to using, to testing, to integrating this into their workflow. But there's also people that are, uh, that are worried that this might take over their job.

[00:20:33] Uh, I will put it bluntly, like that's, that's a pretty, uh, pretty common concern that I get is will this replace me? Will this end up taking my job? And if you're a medical coder, my, my like short and pretty. Like confident answer is no, it's not going to replace you. As I said before, this like being a really, really good coder is an art, and the AI can [00:21:00] augment.

[00:21:01] What you do and how you get to the answer, but it's not going to fully automate how you code an inpatient record, for example. That's very complicated with many different diseases and hundreds of pages of clinical nodes. Uh, it's not there. And I don't think it will be there in the near term. It might be there at some point.

[00:21:20] I don't know how the technology will evolve, but at that point, I think the, the focus and the scope of medical coding might become more supervisory or more critical of the AI and, and more on the training and evaluation side. But I don't think we're anywhere near that, that part yet in the near term, really, I think as a, as a coder, what you should be thinking about is how do I.

[00:21:45] Become better at what I do by being able to use this technology. So in the same way that you might be using chat GPT in your daily life to like plan a trip or, or work through something or [00:22:00] start a new project, um, think of, think of AI in the medical coding space in the same vein. It's going to be able to augment you across the stack eventually of, of what you do, but it's not going to automate you.

[00:22:14] Mallory: Mm-hmm. Do you think there will be less medical coders though if AI's kind of heavily augmenting the existing medical coders

[00:22:24] Dr. Nicola Sahar: there? There might be. There might be less, uh, like traditional medical coders, but I still think, like, at least now, the way that we build our AI is it's, it's very human in the loop and we do require.

[00:22:38] Humans at the end of the day, medical coders to, to work with the AI to build, to train, to evaluate, to supervise. This is in addition to actually completing coding. So there's actually new roles for medical coders for actually building these systems, making sure these systems are accurate. They're not hallucinating, like hallucination is a huge issue for, [00:23:00] for, uh, generative AI now, where we can't really trust what it's saying.

[00:23:04] And so you do need the domain expert. You need a medical coder in this case. To be that final say and to be that authority. And I don't think that's going away like anytime soon. We're not close to solving hallucination in generative models like that. That's a whole other issue, like with the architecture of, of large language models that, that I haven't seen a solution to.

[00:23:27] So as long as that persists, like you're, you're really become even more important to make sure that these systems are, are stable and not hallucinating.

[00:23:38] Mallory: I think that's an excellent way to put it. I know we have many healthcare association listeners who are wanting maybe to do something like A A PC has done, but are certainly concerned, you know, if you were an association of, um.

[00:23:51] Um, I don't know, writers for example, that was trying to roll out a knowledge agent for your association that writers could use that had access to all of your [00:24:00] resources and educational content. That's one thing. If you have, uh, an association of physicians that would like to use your knowledge agent where they put in patient information, that's a whole nother ballgame.

[00:24:10] So I'm curious from like the healthcare data side Yeah. It privacy's essential. Right. And so how did you approach building an AI tool where you could ensure kind of that patient data was secure?

[00:24:24] Dr. Nicola Sahar: Yeah, uh, that's a huge one. And honestly, the short answer is like, it's much harder to do that. And you can't rely on external, uh, external models.

[00:24:34] So you can't rely on API calls to, to open AI, to Google. Google might have like HIPAA compliant models now. In the past, like nobody really did. So you have to build these models. At least the approach that we took was completely private. So on our own private cloud or even on premise. So we've spent many months sometimes with customers deploying to their own servers, like [00:25:00] building and training and testing and evaluating AI on their own bare metal server, which is 10 times harder than, you know, making an API call to open ai.

[00:25:12] But that's the cost of doing. You know, business in a secure and private way in, in healthcare. Uh, and ultimately I think, like you really need to be serious about this because, um, like you, you do need to protect the clinical data. You know, I've heard horror stories these days of like residents or physicians like pasting clinical notes into chat GPT to to write note, like this is a real problem and we do need, we do need like, oversight and, and like secure and private solutions in healthcare, especially on the clinical side, as these tools become like super easy and powerful for physicians to use for writing notes, for example.

[00:25:51] Um, so yeah, security is a huge. Consideration, and you have to put in the time and, and the effort to, to make sure that you're [00:26:00] doing it right.

[00:26:01] Mallory: Mm-hmm. But it is possible. It's feasible. Obviously you've done it,

[00:26:04] Dr. Nicola Sahar: it's totally feasible. A lot of companies have done it and are doing it. Mm-hmm. Like not just in healthcare, banking, insurance, like a lot of.

[00:26:13] Highly regulated industries will require private, sometimes on-premise or securely, like cloud deployed instances of ai. And, uh, and so it's not, it's not a new industry, it's just gonna cost more for you to maintain, you know, your own server, your own cloud instance, versus delegating to an API call, which is mm-hmm.

[00:26:35] Honestly, easy to do.

[00:26:37] Mallory: I wanna, I wanna use the example that you mentioned 'cause I think it's, it's a funny one, a concerning one, but it also circles back to a greater point. So the resident potentially who is copy and pasting patient information into chat, GBT or Claude or whatever, large language model.

[00:26:52] Yeah. I'm sure that's happened many a times. I'm sure that it's still happening even with medical coders. Right. I'm sure as Chatt [00:27:00] rolled out, there may have been some of them that were using it in that way to review their work. I'm curious what you have seen on the A A PC front in terms of. Educating medical coders around artificial intelligence, because I'm sure when you roll out something like Semantic Health to them, there is some education learning curve.

[00:27:19] Yeah. Like you need to know what's possible with AI to understand why you should adopt something like Semantic Health

[00:27:24] Dr. Nicola Sahar: Education, I think is the most important step for, you know, getting a domain expert, like a medical coder to successfully integrate AI into their day to day. And I think education is a few things.

[00:27:38] The first is like. Really just getting everybody on the same page about what AI actually is. There's a lot of like noise about what it is and what it does, but at its core, it's a statistical model like generative ai. At least like the most basic way of putting it is it's a model that uses statistics to generate [00:28:00] text for you based on text that you give it.

[00:28:02] Or data that you give it. And so it's a statistical model and it's not a reasoning model. It's not like a model that can plan or is highly intelligent. It's just a, it, it's basically going to be a pattern matcher. And so a lot of education like goes into that step of like, what actually is it under the hood?

[00:28:23] Because that determines what it can actually do and where it's going to break and how it's going to hallucinate. So the, the more you, you kind of intuit the fact that it's just complete, it's like a very fancy, auto complete at this point. Then the, the more empathy I guess you have for the model for, for what it's doing and when it actually doesn't work and how to actually troubleshoot with it at the end of the day.

[00:28:52] So. Uh, understanding the first principles is step one, understanding like its strengths and its weaknesses is step two, [00:29:00] especially for your workflow, like. In medical coding, it's not going to 100% get every single code and capture every single thing in the document. And sometimes it might code something and explain itself in the wrong way.

[00:29:13] So there's limitations that, uh, that we have to educate people around. And the third piece is just education in general around. Um, the future, like what's your role, how is your role gonna look like? The augmentation versus automation like debate that we just talked about, I think is a huge part of education and, and really aligning people with the fact that like, this is a tool, like in the same way that, you know, calculators came out and you either use it to do math faster or you don't use it and you're, you're still using your pencil to do math.

[00:29:50] Like that's, that's kind of. The way that I look at this, it's like an intelligence upgrade for you, uh, to be able to do more, uh, versus it's [00:30:00] gonna just automate everything away.

[00:30:02] Mallory: Mm-hmm. I feel like associations are just perfectly positioned for this. On the education piece, they're typically, you know, the thought leaders in their industry or profession, so they can provide that education.

[00:30:14] If there are any solutions like Symantec Health that they endorse, they can potentially provide those to members. That's a huge opportunity for revenue generation. Then I love what you said too about thinking ahead, which is something Amit always talks about, not just looking at the landscape. Now that's essential, but when Gen ai, AI in general is moving so quickly, looking maybe a year or two ahead, what does this profession look like?

[00:30:36] What does this industry look like and how can we help our members get there? I would argue there's no better organization positioned to kind of help people through this transition than associations.

[00:30:47] Dr. Nicola Sahar: I agree. That's why we partnered up with, with a, it was a good

[00:30:51] Mallory: move. It was a good move for Seman Health.

[00:30:54] Dr. Nicola Sahar: Yeah, exactly. And I think associations are, are, are really [00:31:00] the educational providers, like we're certifying at the core. Yes. We're certifying and we're providing the, the credentialing and the exams, but, but. Like we're providing education about how things are changing and where things are going at a time when there might be a lot of uncertainty around the technology and around the future, uh, at the end of the day.

[00:31:20] So I think it's, it's very important to stay close to the associations that you're part of, to be involved and to, uh, take part in the courses. Like a PCI think if it already isn't released, uh, has an AI kind of primer course that, uh, we've been working pretty hard on. Uh, I'm not sure if it's out yet or not, but, but these are some of the things that we're focused on.

[00:31:44] Mallory: I'm curious if there are any other, well, there's probably many, but any that you would wanna share on the podcast trend lines that you're seeing with healthcare and generative artificial intelligence or ai, um, that you'd want to speak about? Anything that you're [00:32:00] looking ahead toward? I don't, I mean, I know you're focused on semantic health.

[00:32:03] Any other future startup ideas? I

[00:32:05] Dr. Nicola Sahar: think one of the most interesting things is like how do we use these agents in our day-to-day life to help our own health? Like we focus a lot on physical health and and healthcare, but what about mental health or spiritual health? I think a lot of people are naturally gravitating towards.

[00:32:25] Chat GPT for this type of health. I don't know about you, but I've been using it in in that way. You know, if, if I have a strong emotion or like a tricky situation, I do go to chat GPT for some advice on what to do. And I think I've talked to a lot of people and I've seen some research that shows, you know, this is one of the main use cases of, of chat bots is kind of like our mental health, our, our wellbeing overall.

[00:32:51] Um, so I think that's a very exciting direction. Right now, I don't think chat GPT is equipped properly to do it just with the [00:33:00] hallucination. And then secondly, its tendency to confirm pretty much everything that you say. Uh, both of these things can get pretty dangerous if you're using it for. True mental health support like therapy or counseling through tricky situations.

[00:33:18] So I think we need better solutions that are more grounded in, you know, clinical guidelines, principles and therapeutic principles that have the proper context about people and that are private. And I think that's a really exciting area. Uh, that that can help pretty much everybody because, you know, access to therapy generally is expensive.

[00:33:41] It's limited, I think even if you don't have like a true mental health disorder. Therapy is like working out for your muscles, but, but it's for your mind. Like I think everybody can benefit from, from having these dialogues in general and eventually chat bots might be a [00:34:00] pretty good way of, of facilitating that access to, to people across the stack.

[00:34:05] Mallory: I think that, I don't know if we've ever talked about that on this podcast in our nearly, you know, a hundred episodes, but I think that's so important and yes, I myself have found. Myself going to, I like to use Claude, but yeah, if I have a thought or an emotion, or especially too, I tend to be. Pretty curious and inquisitive.

[00:34:24] So if I'm reading a book about something and I disagree with the point, I'll kind of like voice mode that into Claude, well, I don't really see how that applies to this, and then it'll kind of help me work through that. So I think there's a lot of opportunity there. But I do agree with you if we could have a, a model that was trained on like the, the DSM and maybe like exactly like clinical information around people, uh, and mental health, that could be interesting.

[00:34:46] Is that something you are looking to work on in the future?

[00:34:50] Dr. Nicola Sahar: I don't know.

[00:34:51] Mallory: Yeah, I dunno. I feel like you're having this thought, you've got the experience.

[00:34:55] Dr. Nicola Sahar: I just, I think I like personally resonate with this. Yeah. Because I, like, [00:35:00] I've worked on, like, I've worked in the mental health aspect of healthcare.

[00:35:04] Like I've struggled with stuff too. Like I know people and I think like, and, and I've used chat PT and I can see where it's like falling apart for this use case. But it's a very obvious. Starting point that can, that can help pretty much everybody that has access to chat GPT. Um, but I have no definitive, definitive plans.

[00:35:26] Mallory: You need to do a, a hackathon or something, perhaps another, yeah,

[00:35:30] Dr. Nicola Sahar: maybe a hackathon.

[00:35:32] Mallory: Um, for our association listeners who. Have been surely fascinated by your story by A A P C's story, who would love to do, incorporates a solution like this into their repertoire? Yeah. Who want to be maybe a little bit more like a A PC, more forward-looking based on what you've seen working with this organization, what advice would you have for associations in general that tend to have, you know, smaller teams, limited [00:36:00] technical resources, et cetera?

[00:36:02] Dr. Nicola Sahar: Working with a PC is a great starting point. Uh, we, we do have a lot of knowledge, resources and people that that understand kind of. The coding industry, but also the technical side as well. Um, and it depends on your goals. Like I would say, like, have a, have a, I think the most important thing is like just thinking about your vision and your goals and like where you wanna take things.

[00:36:26] I think that has to come from within, as with, with anything. And once you have a clear idea of. Hey, we want to try out AI or we want to expose our members to more AI or go in this direction. Then you can approach organizations like a PC with, with like a tangible vision. And, and they're more than, we're more than happy to, to kind of explore these things with you at the end of the day.

[00:36:50] But having a clear goal, clear vision, and um, and you don't have to do it all, especially with ai, like, I don't think, I don't think that there are. [00:37:00] Like experts running around that know AI everywhere. There's a lot of people that are using it and are building expertise, but like true expertise from the technical level, from the practical level, that's still very rare and organized, like larger associations like a PC, you know, at Symantec, that's all we've been doing for the last.

[00:37:21] Eight year, or not eight years, six years now. Yeah. Or, or so. So, so reach out to people like that. We're not the only ones. There's other, or associations, other companies, researchers at universities. There's a whole bunch of people that, that really know what they're doing and that are more grounded and less kind of hyped about the, the tech.

[00:37:42] They're still excited, but not, not hyping it up unreasonably. I think those are the types of people to, to seek out for like a grounded perspective.

[00:37:51] Mallory: Yep. And when you encounter people in your personal or professional life who are a bit more cautious about artificial intelligence, which is [00:38:00] fair or even perhaps against it.

[00:38:02] Totally. What do you typically tell them?

[00:38:05] Dr. Nicola Sahar: I mean, I'm also cautious about it from the point of view that like it's a statistical pattern matching machine now, and we're applying a lot of like. Abstraction to it and, and, and kind of like meaning to it and anthropomorphizing it, to be honest, and I don't think it's as smart as, as we think it is right now, it's gonna get better, but like there's a certain level of caution you have to take when you do this because you start applying it in different ways that maybe it shouldn't be applied with some level of bias or blindness about, its, its actual like.

[00:38:40] Effectiveness, effectiveness and reliability. So I think you should always be cautious about ai, but I, I also think it's, it's gonna be the most impactful technology ever. Like in our species, in our sp history of our species, uh, if we build it out properly, [00:39:00] uh, because it's an intelligence, like level up pretty much for everybody.

[00:39:05] Uh, and, and like it's going to be exponential at some point. So it's going to enable a lot of other fundamental discoveries over time as well as you and your daily life to, to, to just be much higher quality. Uh, that's the vision I hope we go down. But, um, but we have to be cautious because like this technology is pre, is pretty powerful now and it can get a lot more powerful.

[00:39:31] So we have to build it in a responsible, ethical way, make sure that it's aligned with us and. And really just focus it in on things that matter. Uh, uh, at the end of the day, like healthcare for, for me is, is the thing that matters. So that's why I'm, I'm focused on that space.

[00:39:50] Mallory: I think that is a very healthy outlook, being cautiously optimistic.

[00:39:54] We like to say that we are both of those things on, uh, the Sidecar Sync Podcast. Nick, I've gotta say [00:40:00] thank you so much for joining us today. This was a really insightful conversation for me, for our listeners I know, and so we really appreciate your time and we look forward to seeing what's next for Semantic Health on the horizon.

[00:40:13] Thank

[00:40:13] Dr. Nicola Sahar: you so much, Mallory. It's. Been a pleasure. It's been super fun, and I hope you guys enjoyed.

[00:40:20] Intro: Thanks for tuning into the Sidecar Sync Podcast. If you want to dive deeper into anything mentioned in this episode, please check out the links in our show notes, and if you're looking for more in-depth AI education for you, your entire team, or your members, head to sidecar ai.