Health Information Exchanges with Benjamin Rudeen

Exploring AI in Mental Health Care

Benjamin Rudeen: And so there’s a lot of work we’re putting into in terms of how does the patient interrogate how they feel and why they feel that way, using some cool AI tools. And then it’s not just the patient, right? It’s the clinician. We’re thinking about how do we help the clinician get as much information and insight as possible with taking as little time as possible?

Because again. Especially in mental health, the clinician time is very valuable. Clinicians are way overworked. And so we don’t want to add time to get these insights. And so how could we use AI to summarize certain things or point out trends or flag certain patients?

Introduction

Bruce Bassi: Welcome to the future of psychiatry podcast where we explore novel technology and new innovations in mental health. I’m your host. Dr Bassi an addiction physician and biomedical engineer today.

We are with Ben Rudine who is the operations lead At headlamp headlamp makes it easy for clinicians and their patients to access their universal medical records Through a multitude of health information exchanges while also providing advanced AI tools to support more personalized Health insights and decision making Welcome Ben

Benjamin Rudeen: Thank you for having me, Dr. Bassey.

Bruce Bassi: Do you mind telling the audience about how you kind of heard of the podcast and what your kids think of it

Benjamin Rudeen: Yeah, no, I’d be happy to. Yeah, I try to listen to a lot of different podcasts in the car driving. I have a two and a half year old and a one year old, and so we spend a lot of time going back and forth to daycare. And rather than get stuck in top 40 radio, I try to listen to podcasts to keep my brain running.

And since I’m in the mental health space, I listened to a lot of different mental health podcasts and came across yours. And the 1 year old was very vocal about it. who knows what she understood? Probably nothing, but, that was a good sort of sign of approval. So, that’s sort of how I stumbled across and then.

Here we are today. The two year old, two and a half year old had a lot less to say, but I won’t hold that against

Bruce Bassi: They’re going to be so proud that you’re on a, such a popular podcast show now and you’re going to be famous.

Benjamin Rudeen: Of course. Yeah.

Bruce Bassi: Do you want to introduce yourself a little bit in terms of how you got involved in this, technology and, why you’re passionate about it?

Benjamin Rudeen: Yeah, of course. I’ve always had an interest in science. I was certain I would do a PhD when I was younger. Um, when I was in high school, I saw myself getting my undergraduate degree and going on to get a  PhD. And was actually went in as a chemistry major. I actually got a degree in chemistry, but through the coursework and the lab work, I became very disillusioned with pure chemistry.

a lot of my peers were doing catalyst creations or optimizations, and I found it very sort of very boring and didn’t seem to have for me to have the impact that I wanted to have. I know that the catalysts can’t have a large impact, but, I found myself more and more increasingly interested in the brain and initially it was disorders.

my mom has multiple sclerosis. so I’ve thought a lot about that and I got really interested in Alzheimer’s and Parkinson’s. I did a summer research in a Parkinson’s lab with Dr. Kurt Freed. He did some of the early stem cell implants for Parkinson’s patients. And so we actually got to hold one of the brains of one of his patients.

He did a stem cell transplant for and so became very interested in the brain, started taking neurobiology and learning and memory class with a psychology. So in the psychology department. And still thought I’d do a PhD, but wanted to do it in the, that space, but became secondarily disillusioned with the gap between research and  clinical and actual patient care.

The senior product I did was very interesting neuroscience work, but it was work that maybe would help a patient 20 years down the line. And to me, I need something that’s a little bit soon than that. I took what I thought was going to be a gap year after college. And I told myself, I take a gap year and then go to get a PhD.

And thought I would look into working in the biotech space and for my gap year we found a way to Chicago and then I found my way to a biotech startup in Chicago they started in the oncology space, but when I was there, moved into the mental health and neuropsych space. And so I really got sort of cut my teeth on operations and strategy at a biotech startup and actually got to build out the neuropsych product alongside the team and became just fascinated with the new technologies growing around different treatments, different genetics, different biomarkers, the search for precision psychiatry became my obsession.

And I was lucky enough at that biotech to really explore that and then got lucky enough to get connected with the co founder of headlamp and join where we both have similarly very passionate about precision  psychiatry and moving towards a more data driven biomarker fueled world of psychiatry.

And so that’s how I came to where I am today.

Track 1: Nice.

The Complex World of Health Information Exchanges

Track 1: I know that headlamp operates basically around health information exchanges and believe it or not, somebody who’s in mental health, and also interest in technology. actually, I’m embarrassed to say, I only learned of health information exchanges like a year ago. And I look them up and I’m like, where, have I been under a rock?

I should have known that there was some magic behind the curtain you know, I probably remember back in 2014 when I was in an epic chart and you can click care everywhere or something like that and you can see other. records and I just, thought that there was like some sort of special agreement among hospital systems in New York City at the time.

lo and behold, little did I know that there was actually this huge, massive movement also from the government to try to encourage this too, into creating these exchanges. can you tell somebody who may not know much of about what these are, about a what they are and be why they’re important.

Like, why does the government, mainly like state funding  think that they’re so important to help, grow right now?

Benjamin Rudeen: Right?

The Challenges and Potential of Health Information Exchanges

Benjamin Rudeen: Yeah, health information exchanges are

almost spoken wheel there. There aggregated health information data from clinics and hospitals across the country, and it’s a really complicated network because there’s state level health information exchanges, there are larger national health information exchanges that correspond to the smaller ones, and there are some that are independent, and it’s a very complicated web, and The idea behind them was very interesting and the right idea, which was increasing interoperability and actually just making continuity of care really possible because it still shocks me to this day, how hard it is for if I go see a new doctor when I moved to Chicago for them to know my health history, it required a lot of work on their part and my part.

And even then it was not really good information. And so the idea behind them was great. And it came along with the meaningful use. Sort of legislation and just trying to get more people to use electronic health records in general, which have been, has been great. I mean, we can get into EHRs at different time and they’re pros and cons, but the idea behind them was  exactly what we need, which is to fuel a single place where you can go and find.

All the medical records of yourself and others and really divine insights from them, whether that’s just access or also really deep diving into them to see what trends or what patterns emerge, but they really do seem like a best kept secret because no one really seems to make use of them and they are.

It feels like a very complicated system because they’ve been like. Layered on top of each other for so long. Thank And so they have a lot of really interesting information in them, but there’s a lot of challenges that go along with using the health information exchanges and accessing them that we’ve come across.

And so they are great in theory and idea of let’s get all this data in 1 place. We can really make use of it. But the practice of it is as much of health care is shown to be the practice is much harder than the theory.

Track 1: Right. Yeah. In preparing for this conversation with you, I was listening to a lecture on HIEs and. They were talking about how many there are in just New York State alone, and I was like, oh, no It was really painful to listen to that. There’s one for Buffalo and one for Rochester and one for every other city in New York and then lower New York and I’m like, we need an exchange just for the exchanges or a network just to regulate the exchanges and it’s like probably thing that they were trying to solve is the interconnectedness of information has now become another like secondary issue on top of that, like an another It’s the nth order of an issue for the exchanges themselves to talk to each other.

it was kind of sad in a way, actually. I felt a little bit sad hearing about that because in a way they tried to solve a problem, but created a different form of the problem they were trying to solve.

Benjamin Rudeen: Exactly. That’s a great way to put it. They’ve created so much. It’s all the data is all in 1 place, but it’s just noisier than ever. it’s almost just as bad as not having the data in 1 place. And so I love the

idea of it,

but they’re challenging to work with truly.

Track 1: Yeah, and another thing they talked about was, next step, even if all of the data is in a perfect world, in this aggregate in one repository where somebody can access, then it’s now making it useful for that clinician. how do I sift through if it was just PDFs, like. How do I sift through a thousand pdf  pages, sufficiently or like efficiently and find some other sort of very narrow clinical question that I’m trying to ask in that encounter who regulates them at this point or is there like an agency that can help Make it a little bit more consistent among different HIEs, or what’s the beat on the street in terms of like how things are going to start improving from here?

HIEs: Incoming Improvements

Benjamin Rudeen: Yeah, it’s a great question. I wish there was a single entity that could step in and clean it up. But mean, in health care, that’s never the case. I do think, though, that we are on the precipice of a lot of people really making good use of this data. And I think the more people that use it, the more people that will cry for standards and just better practices.

I honestly think for the first 10, 15 years HIEs existed, they really were just fueling some small back end processes and not really doing a lot of front end clinical impactful care. They were Doing a lot of important things, but not really people were not really digging in and trying to really interrogate them for answers.

And I think we’re at that point now, where there are a lot of really interesting companies coming around that are trying to use these large data repositories, data sets to really make meaning and drive change in health care. And I think we’re going to come to a point where we’re going to have some sort of coalition of these companies come together to say, look, we need this data to talk with each other.

We need this data to make sense and all harmonize. So let’s work towards that model. And there’s a few models that are coming together for a standard data model that would ideally help to harmonize. But additionally, when you have, especially mental health, such a fragmented EHR market to get all of those players to all sort of coordinate with the same model can be challenging as well.

So another part of it might just be, we might see some M and A in the EHR space that might sort of make it a less crowded playing field that might make it a little easier to make it a little bit interconnected. But right now it’s like you said, it’s a mix of pdfs and actual sort of data fields.

And then there’s no standard on whether you use ICD 9, ICD 10 or RxNorm or MedDRA. And so they all end up meaning the same thing, but slightly differently. I actually think LLMs might be a really interesting use case here because if you were to give them two different HIEs that had slightly different ways of entering information.

It could probably do a lot of interesting harmonization. I saw there was a project done recently that was doing that for clinical rating scales, so you could feed in a ham D and back depression inventory, and it could tell you which items most sort of closely aligned with each other. And so I think we might start with LLMs to really get a better sense of how do we start to harmonize the data.

So that also might be something that comes up soon with LLMs. There might be someone who takes on that heroic task. But as of now, it’s really, I think people are sifting through what’s there and we’ll start to sort of see that need and that call for action, I think, in the next few years, hopefully.

Bruce Bassi: I think the ability to be able to access records a very easy and efficient way, for my patients will. Of course lead to much better care, better communication among clinicians, and I think that kind of goes without saying. And I think there’s tons and tons of advantages, but do, you know, other than privacy, and privacy is like a huge topic and making sure you appropriately get consent from the patient and whatnot, not sell people’s data, but.

Are there any other, like, unforeseen negative consequences to having these from like a clinical perspective that, maybe aren’t talked about as much,

Benjamin Rudeen: Yeah, it’s a good question. I think one thing that we’ve come across is we know that the EHR data isn’t perfect. Right? Like we’ve all seen some of these progress notes where there’s one med that’s been duplicated 100 times or a diagnosis that’s no longer active and propagated over and over

Bruce Bassi: or wrong.

Benjamin Rudeen: Right? and so there’s a desire. I think of a lot of these people who are working with the data to want to improve it and sort of inject other sources to make it more accurate. But then you get to the question of how do you attribute? Where the data is coming from and how to trust certain data sources.

like if you pull in patient reported and pharmacy, like PBO organizations, how do you reconcile the medications in an EHR with the medications in a pharmacy record? course, you probably take the pharmacy record, but you start getting in this rabbit hole of the data is better, but we’ve introduced so many more sources.

It’s hard to know sort of what’s trustworthy and what’s not. And a lot of times there’s a lot of data that only the patient can really give you, like it’s not in health encounter data, but then is it valuable to have that data next to objective health encounter data? Yes, but then also you’re introducing objective data next to subjective data, and that introduces a lot of questions too.

And so it’s something we think a lot about in terms of, it’d be great to take this health record and make it better, augment it, add to it, really. Massage it and make it better, but then what are we introducing into that that would might influence or impact downstream analysis or downstream issues. And so that’s 1 piece, and then another piece is just who owns the data and also just the way these things talk to each other, because.

when you go to a health information exchange and pull information down. Something might be wrong, but then is it who’s responsible to, like, tell them it’s wrong, but it’s not theirs. They just pulled it from the source. So then correcting that information becomes kind of. Hairy, and then you don’t really know who owns it at that point.

Bruce Bassi: whose risk is associated with it too?

Benjamin Rudeen: right, yeah. And so there’s a lot of cooks in the kitchen when you start to actually integrate these HIEs and try to add information to them. makes it challenging. I would love it to then grow to a larger source where it becomes one true source of truth that is mutable, that can be sort of updated.

But right now it’s a bunch of one way streets, we get the data in the HIE that we can use, say, in Headlamp, but that’s come down multiple one way streets, and so what we have might be wrong, but there’s no way to actually Propagate that change to make sure that it’s accurate. that can be really challenging.

Bruce Bassi: Yeah. that’s really interesting because as you were talking, I was thinking maybe these are all just really good problems to have. It’s like, say your plane is late by an hour, still are flying 550 miles an hour through the air, And it beats driving for four days. But, having some data is better than no data because otherwise We’d be, you know, where we were five years ago, which would be having nothing, but it’s possible that, you know, there’s could be instances, maybe cases where having the wrong diagnosis or the wrong medication that somebody maybe just like didn’t have the time to reconcile completely because everyone is so rushed.

there might be instances where they’re like blaming another health care organization for the wrong information. it sounds like kind of far off, but I don’t think it’s like totally  impossible to eventually happen. But, I mean, it’s probably not any different than just getting the records from a fax machine anyway.

but you’re right though, it seems like since there are so many sources kind of coming into the same place, it probably, it just makes it a little bit more confusing than just getting a singular fax.

Benjamin Rudeen: Yeah. And so what we like to think of is just the first step you can just do is make it transparent, right? Just know where the things come from. And then we’ll work it out later. A better way sort of to circumvent that. But. Yeah, I mean, you’re right. I, would I rather have someone fax me a thousand pages of records or get a data dump that I can at least sift through on a computer?

I’d probably prefer the computer one. yeah, I mean, in psychiatry, sometimes I feel like it is choosing between a sort of good and a slightly better. And, I mean, I’ll take the slightly better, and I think it’s always just about building towards the best.

Bruce Bassi: Yeah. And it also reduces the time spent, both administratively and also waiting for the records to come back too.

Benjamin Rudeen: Exactly.

Yeah. And I think there’s also just the intangible benefit for it being available in public, right? Like, I think we’ve seen that when the patient has access to their history, they’re more engaged with their care. we know the clinical value of having more than just sort of self  report data, but I think just to that point, just having it be out there, not sort of hidden away or making it hard for patients to get or hard for clinicians to get.

We’ll start to bring up more of these questions of like, how can we make this data the most useful for patients, for clinicians, for payers, for everyone in the ecosystem. I think, better to have it present and there and really dig into it, then continue to sort of leave it behind the curtain.

Government Initiative

Bruce Bassi: Yeah, in also preparing for this conversation, it sounded to me, and correct me if I’m wrong, that much of the initiative had been taken on behalf of the government to try to start this. why do you think that was? Maybe just because, you know, it could never have been privatized from the start. also Is it not inherently, monetizable, like from a private company if that were to happen?

I’m talking like completely hypothetically here. But, I’m kind of curious, more than anything, like why the government, because they don’t really do a whole lot themselves, left up to their own devices, so they really need to see a benefit for something like this, or there’s really passionate people behind it trying to push that legislation through.

So, I mean, there must be really Strong advantages from a public health perspective for why this is advantageous for them, maybe reducing Medicare costs or what have you. I’m just hypothesizing, but can you speak to that at all?

Benjamin Rudeen: Yeah, so I think. What drives it, honestly, is what drives a lot of innovation in healthcare, unfortunately, which is billing. I think that’s what’s entirely driven how EHRs are built now. That’s what’s driven, unfortunately, how diagnoses are listed a lot of times, like ICD 10 codes and the way we use those, they’re not truly, they’re meant for billing.

They’re not truly representative of that patient. Right? So the HIEs really came about as part of the mission to get, All of hospital, most hospitals across the country to use EHRs. And so before the meaningful use action, high tech, I think only 10 percent of hospitals were using the EHR. And so that mission, now it’s ubiquitous.

It’s hard to find a clinic or a practice that doesn’t use an EHR. And so that need to have a clean billing for Medicare and Medicaid, but also just generally to have that more streamlined billing process, I think is what drove it. And I think there’s a lot of good that kind of came from that. But then I do think because it’s built around billing, it’s something that I’ve seen as I’ve dug more and more into healthcare and more and more into medical records and.

Everything in that realm is that when you build things around billing, it gets things done because money talks, but then you, you build yourself into a whole of seeing it in a very specific light and having a very specific use case that’s not in the best interest of the patient. It’s in the best interest of sort of the insurance process.

it’s good and bad because it spurred it along. it absolutely did its job and getting it sort of getting the HRS adopted, getting these exchanges set up. but it was built, I think, with the wrong. Well, it was built specifically for billing, but that was not the right way to be built to enable a lot of these better clinical use cases that we’re talking about today.

Bruce Bassi: Yeah. that makes sense and seems like an unfortunate, sad reality of the way things are in health care. lot of the time.

Personalized Mental Health Care with Headlamp

Bruce Bassi: So say there’s a listener out there who’s super interested, they’re sold, they think this is super beneficial, they want to get, um, more involved with headlamp and what you all are doing, what can they expect, how does that work, and where do they go to get more information?

Benjamin Rudeen: if there’s clinicians out there who want to learn more about Headlamp

health, they can visit our website and they can schedule time to talk with me or someone else at the company from the website. we try to make it as easy as possible. And to get you get the clinician into our platform and that you enrolling a patient takes seconds.

And then that data on that patient that we’ve spent a year and a half digging into and cleaning up and making useful that data is available to the clinician within minutes sometimes. And then we have a company mobile app where the patient actually gets invited. They get access to their medical record.

So they can have a say and really make sure that it’s really telling their full story. But then they also start to monitor their mood, their behavior over time. In a really personalized way and really dig into why do I feel this way and what can I really start to do or focus on to maybe feel a little bit better?

it’s a really cool sort of, I like it ’cause I think a lot of mental health apps, either it’s an EHR and it’s fully like in the hands of the clinician, or it’s a wellness app that sits outside of sort of the patient clinician relationship. And what I like is we built something that really is grounded in the therapeutic relationship, which I think really helps.

And it just shares data among everyone to make sure that we can make better, decisions. And so if someone’s interested as a  clinician, we’re happy to share more and talk about how we can help the clinician and their patients. our website has information for patients too, because we’re just as, or if not more interested how do we help patients feel better and just understand more about why they feel the way they do.

It’s a really complicated question. Like I can maybe start to articulate. I feel sad. Um, But why I feel sad in a given day could be a lot of different reasons. And so building for the patient to build some awareness around that, is another mission that we’re really excited about. And so we have some information for patients on our website as well.

Bruce Bassi: Provides advanced AI tools to support more personalized health insights.

AI for Deeper Health Insights

Bruce Bassi: Talk to me about how Headlamp is using AI to give more patient health insights and decision making.

Benjamin Rudeen: I’ll give a few examples.

and one is, I think every patient who’s gone through psychiatry or therapy has heard, you should eat right. You should exercise. You should sleep more. there’s just like the sort of common lifestyle, recommendations and I don’t know anyone who can do all of that, I mean, I have two young children.

I certainly can’t sleep more. I also know that for some people, sleep really doesn’t seem to affect them. And for some people, it’s like you could not talk with them if they’ve had less than six hours of sleep. there’s starting to be this, you can start to see where it’s not all these variables are equal for everyone.

And it really does depend on that person, where they are, their condition. And so what we have some tools around is based on how you tell us how you’re feeling and how you’re behaving, how you’re acting on the world, your behaviors, your sort of lifestyle, we start to tell you, Hey, there seems to be some interplay between your socialization and your happiness, and maybe it’s actually too much socialization makes you unhappy.

And so we start to really tease out sort of this higher level because you can just keep telling us you’re unhappy, unhappy, but now we can sort of step up into a higher level to say. There seems to be something between socialization and your mood. How can you, with your clinician, dig in to really explain, Is it social anxiety?

Is it just that you’re an introvert and you just need to spend more time sort of finding time alone to get energy back, or is it just the type of people you’re surrounded? Maybe you just need to get rid of some friends. there’s a lot about these variables that are so personal that we sort of want to  start teasing that out and to go even a step further with your past medical information that we’re able to capture with other information, just like where your phone is located.

We know the weather. we can start to do some really interesting things to say, like if you tell me I’m sad, is that because I’m in a depressive episode? Or is that because it’s been rainy for five days and I haven’t gotten outside? And you can really start to notice these patterns in yourself that, of course, I could say it, like, of course, I’ll feel less happy if it’s cloudy for five days, but it doesn’t really cement in my mind unless I really see that in the sort of cold, hard facts,

and you can really start to just be more aware of what around me seems to be contributing to how I feel and how, what can I do to control that, or at least, you Monitor that to make myself sort of feel a little bit better.

And so there’s a lot of work we’re putting into in terms of how does the patient interrogate how they feel and why they feel that way, using some cool AI tools. And then it’s not just the patient, right? It’s the clinician. We’re thinking about how do we help the clinician get as much information and insight as possible with taking as little time as possible?

Because again. Especially in mental health, the clinician time is very valuable. Clinicians are way overworked. And so we don’t want to  add time to get these insights. And so how could we use AI to summarize certain things or point out trends or flag certain patients? Maybe there’s a patient who could be eligible for TMS, or maybe there’s a patient who is a good candidate for genetic testing.

we have data to tell you about these patients, so why not do some work to sort of help administratively to say, I don’t have to dig in to find this information I can have sort of delivered to me. And so we’re really excited about the opportunities that AI has, especially for both the clinician and the patient user to really drive just to make these sort of higher level decisions and higher level sort of patterns.

The Future of Mental Health and AI

Bruce Bassi: It’s really interesting to think about, how to start to extract better information from the patient as they’re kind of engaging in a, delivery of such because often when traditional method is to give them a standardized questionnaire and that standardized questionnaire is by definition not going to keep changing.

And it’s not going to look into other elements of that person’s life and it’s helpful in that you have one common comparator that you can use against a larger population. And then it’s, not so helpful in the situations that you’re describing. Where I’ll give you an example, we were at a team meeting today, actually, and I saw in one of my clinicians notes that said that this person’s mood had fluctuated every four to six weeks.

that like just stood out to me because that’s kind of odd to think about. okay, maybe the first episode and the second episode were four to six weeks apart, but then. The third episode being another four to six weeks apart after that, those other two, like that seems like more than just chance, you know, there’s got to be like something related that maybe isn’t, completely obvious to the patient for that’s happening, like you said, the weather or some other work related factor that’s cycling on a, like, frequency of every four weeks.

or some interaction with a family member that’s happening every four to six weeks. and so, computers are very good at something like that to like eliminate bias and just to be able to look at all elements from a person’s life. how does Headlamp go about doing that, like, to start to collect information beyond the standardized PHQ 9?

Like, how do you approach that problem and how is it going to be presented to the patient?

Benjamin Rudeen: Yeah, it’s a great, great question. What we’re interested in is having that sort of data really helps us make smarter sort of suggestions or insights for the patient. But we also know that most mental health apps have really low patient engagement and retention. like 3 or 4 percent retention rates over 30 days.

And so I think there’s burnout among patients just getting asked a ton of stuff. And so we think about how can we engage the patient to give us information in a way that they want to keep interacting with us. And also they see the value of giving us that information. And so there’s two ways we kind of think about that, which is one where we do exactly.

We give them information back when they give us information. I think a lot of these apps are like, just keep telling me your PHQ nine over and over. And that’s like, okay, I get my scores, but like, that’s not cool. So rather than just continually asking me to do these surveys over and over, these assessments over and over, we ask something of  them, then we give them something back like, oh, we noticed this correlation or this possible correlation.

Have you thought about this? Considered why. and so something I think would be cool is like, Hey, like we noticed it’s going to be sunny this weekend. Have you thought about going to a park nearby? there’s just like little things we can start to do that are so much more sort of they give value to the patient.

So they want to come back and give us value. And so we just, it’s a two way street. You can’t treat them as if they’re just dispensing numbers or dispensing data. They came to the app to crave something to feel better. And so we think a lot about how do we ask information of them in a way that we can return interesting insights back to them.

The other thing is also, there’s a lot of ways I am especially interested in how do we measure cognition or mood or. Other factors, through proxy measures. And so for instance, on our app, you can play a game that’s modeled after the Stroop test. And so you play the game where you’re, you have a color that is written and it’s in a different color, and you have to tap the box of the written color, and you have 15 seconds to do as many as you can.

And so it’s a game that’s engaging and fun, but you’re also testing cognition through a sort of the Stroop test, sort of. And so there’s a lot of really cool ways, I think, to get insight from the patient without asking them for something that’s tedious or takes a lot of effort just to seem boring to them.

And you can do it through games, you can do it through Apple Health integrations, where you’re getting the data passively, they’re not even or even just by there’s really cool research on figuring out cognition through how keystrokes work or I think I just came across a really cool paper around the entropy of unlocking your iPhone or your phone correlates to depression symptoms. you can start to even think about how often am I unlocking my phone?

Am I doing it frequently, like, consistently, or is a lot clustered in one time or another? There’s another app, I think, out of Dartmouth that’s being developed that uses when you unlock your phone, it takes pictures of your face and starts to determine if you have depression based on your facial features and your background.

And so it sounds a little big brother in a little way, but there’s also some really cool ways, I think, to work with a patient to say, You’re going to engage with the app, it’s going to be interesting, but we’re also going to collect some information that might surprise you and we’ll give you that information back.

It’s that key. We’re not ever just hoarding it and we’re not selling it to advertisers either. We’re collecting it to give it to the patient and then to use it in further research in an anonymized way. And so when we build around the idea of being engaging to the patient and giving the patient value, that’s our North Star for our patient app to make sure that we can get.

Good data that can give us better insights than just sort of the standard, assessments that we’ve usually gotten that give you a good benchmark, but leave a little wanting in terms of that personalized sort of tracking.

Personal Aspects

Bruce Bassi: That seems like a pretty complex and challenging undertaking to try to do. I think a lot of Big companies are trying to work on that too. I think it’s almost like a trade off. You know, at one end of the spectrum is those standardized tests that are static. And then on the very far end is the super customizable and specific, of data that you’re describing.

Which, you’re getting too specific into that person’s life, it could feel probably downright weird for that person to, think that, I’m now getting recommendations as to, like, very personal aspects of my life from this, like, system that is monitoring me. And so you have to find that sweet spot in the middle it’s not too standardized, but it’s also more engaging than that, but also not, to the  point of just being completely invasive into your life.

that seems like a pretty challenging technical dilemma for you guys.

Benjamin Rudeen: It’s it’s very challenging. And you’ve also think about if you’re trying to build a data set for research, you can’t have all the patients have super specific data that’s being collected because then, like, you don’t have a data set to use because each

patient has a different set of data points. And so it’s something we think a lot about and we’re lucky to have a lot of really great customers and advisors and clinicians that work with us to help us answer that exact question.

it’s right. Cause I would love personally super specific data on everyone, but then the research side of me is like, well, what do you do with that? And if you’re not getting the

same measure, every single patient, your end is one every time, and that’s not the value of big data. And so it’s why we have, standard assessments in our app for everyone, but then we also have the personalized.

And so you kind of write, you have one foot in each side to really see. The best of both worlds and ideally not the worst of both, it’s a very tough question.

Turning Solutions into Bigger Problems

Track 1: It’s interesting to think about too. so I’m really excited to kind of see where that goes and, where Headlamp ends up in, trying to solve that problem.

You know, one interesting thing that we keep talking about is, that we’re trying to solve problems ultimately, like the EHR system itself, if I  were around back then when like the first EHR came about, I would be like, wow, this is amazing. Like, I think this is going to be so cool.

It’s going to be so fast. I can probably save a lot of time here and not have to write things out, which just takes a lot longer. And then, like, I would have never imagined where things would be ten years from then, you know, of like how you have to basically write your note for ten different audiences and like have multiple drop downs and multiple check boxes and menus within menus of like trying to figure out where things go and then dealing with bugs and like things crashing and stuff like that.

And so You like save time, but then like as the tumbleweed just like kept growing bigger and bigger it’s like became like a bigger thing in itself and maybe we can end on the topic of AI based note writing because it’s kind of similar And maybe it has like a similar theme where ultimate goal is to save time benefit the clinicians and downstream the patients and stuff.

but then I experimented with a couple of them and I was like, trying, I was ended up spending cumulatively like more time trying to tinker with the engine and like figuring out like the

best optimal scenario  for it. And then, merging my note with its note and making that seem well integrated.

 we’re figuring, like, how to, like, reconstruct our consent forms for that, and, like, we have to get established our BAA and, like, contract with them, and, like, it just became, like, a bigger thing than just typing the note, ultimately, so I don’t know if that, like, triggers any thoughts for, how Headlamp’s mission how you want to think of that as a backdrop for what you’re doing.

Benjamin Rudeen: yeah I know it’s something I think a lot about and I think I see it all over my LinkedIn. I see AI is doing this and I just think cool but let’s see how that actually works. But I think. It all comes down to what is the problem you’re actually trying to solve, and are you solving that, or are you solving a different problem?

a lot of companies, especially when they see the promise of AI, they think they can solve this big problem, but when you actually talk to the clinician and see their workflow, that’s not the problem they’re having. They’re having a different problem, or you solve one problem, but you introduce another one, right?

And so I think it really comes down to working with clinicians and listening to them and thinking, what is the actual problem we’re trying to solve here? And let’s not bite off more than that right now. So for

me, like, AI note taking is interesting, but I think there’s parts of the note where it makes sense, and there’s parts of the note where it absolutely does not make sense.

And so for me, I just think a perfect marriage here is have a note, a system like Headlamp, where we could generate your HPI, right? Like you could generate like your quick summary of past history of present illness, right? You could say they’ve had these diagnoses, they’ve taken these medications.

Populate something like that and then leave the impression and the actual discussion and subjective part to the clinician. I think there’s a lot of promise with AI, like in the note taking space and other places, but I really think we need to boil down the question to is the issue that. It just takes too long to write the entire note or is the issue that there are parts of the note that you’re doing the same every time that don’t need to be manual and there’s those could be automated or is it that there’s better ways to prompt certain things or you as a clinician can put in X information and generate the structure could be generated by AI there’s just like I think that if you really dug in there are some things that I could really help with saying the note taking space.

But I think a lot of people get sort of blinded by the shininess of replace all note taking with AI, like make it an AI scribe. And that’s hard for me to imagine that fully working out. But I think when you really stay close to the customer and product market fit, you can find problems that you can solve that actually help them without introducing other nuances that make things even harder.

And so that’s, we, as part of our first 20 customers, every single one of them, we asked. Can we just talk with you every other week, because we just want to know what you’re facing in the clinic today. And when we built our, we took months before we even came across what we were building, just talking to clinicians.

And so that’s always something I found to be very important and is just make sure you’re solving an actual problem, not creating a problem to solve. because I think, especially in tech, that’s a very shiny thing to do is to create a problem to solve versus. actually help a user where they are today.

Bruce: Yeah.

Closure

Bruce: Ben, this has been a super fun conversation. Really interesting, too. I appreciate all these, the insight that you’ve given me and all these, additional things to think about, in terms of where things are headed in terms of health information exchanges. thanks for talking to us about headlamp, too.

And if any listeners are interested in Learning more about Headlamp, we’ll put the description and website in the show  notes. Thank you so much, Ben.

Benjamin Rudeen: Yeah. Thank you for having me. This has been great.