William Crawford – Head of Advisory and CTO, Newfire Global Partners

In this episode of HealthBizTalk, host Tony Trenkle sits down with Will Crawford, Head of Advisory Services at New Fire Global Partners and former CTO of Medically Home, to explore how digital health innovation is reshaping care delivery. Drawing from over 25 years of experience across clinical trials, EMRs, wearable tech, and home health, Will offers deep insights into designing patient-centric technologies, navigating interoperability hurdles, and rethinking workflows through automation. From advancing remote care to unpacking the role of AI in healthcare, this conversation delivers a compelling look at the past, present, and future of health IT. Don’t miss this dynamic discussion filled with strategy, design, and transformation.

00:00:00 Intro
Welcome to Health Biz Talk, the industry’s leading podcast that brings you today is top innovators and leading voices in healthcare, technology, business and policy and here’s your host, Tony Trenkle, former CMSCIO and health IT industry leader.

00:00:16 Tony
OK. Hello, Will. How are you? I want to introduce everybody to Will Crawford and Will is the Head of Advisory Services at New Fire Global Partners, where he and his team partner with digital health companies of all sizes to execute on their product Rd maps and solve hard health problems, hard healthcare problems, I should say.

He’s also a venture part capital partner with Spring Tide and Early stage funded invests in digital help and and deep health tech investors, as well as an advisor to health tech companies through New Fire and independently. Over the last 25 years, he’s been a product engineering and operating leader at companies range from clinical trial management to Ehrs to wearable tech, which he did at Fitbit as AVP of engineering. And before joining New Fire, he spent four years as the Chief technology officer of Medically Home, where he led to design, development and implementation of the company’s technology platform for enabling high acuity care in the home. And Will and I first met each other about almost 20 years ago. Believe it or not, Will when he came down to.

00:01:39 Will
The math The math cheques.

00:01:41 Tony
Yeah, at that. When he came down and spent a short time at CMS and we worked on the, you know, consumer focused healthcare records and a number of other things while he spent a short time there. And Will’s going on to bigger and better things since then. So we’ll have a think. We’ll have a good discussion over the next 45 to 60 minutes. So Will, I guess I’ll start the way I usually do and it ask you to talk a little bit about the how you got here, your background order, some of the key points, that key Inflexion points that got you to where you are today and just a little bit more. You don’t have to get into your hobbies or anything else like that. We’ll discuss that later, but it’s from a career standpoint.

00:02:32 Will
Sounds sounds good and and Tony, it’s great to great to be on the on the podcast with you. And you know, I’ve really enjoyed how you and I have been able to stay stay connected at conferences and elsewhere over the last. I can’t believe it’s almost too decade, you know, if the you know, closer to the beginning of my career than than to now, which is is actually very unsettling now that I think about it. So, yeah, but how I got into this, it was almost an, an accident. You know, I, you know, my, my current business partner likes to say that, you know, life is what happens when you’re sort of making other plans. And so, you know, I was back actually in high school and I was picking my younger sister up from elementary school.

This is not how the story usually starts. And I was looking for it was, it’s in Boston. It was, it was summer, it was early summer. It’s hot. I was looking for a job that involved computers and air conditioning and ideally not lifting up the air conditioners. And one of her friend’s mothers worked in the IT department at Children’s Hospital here in Boston. And he was friends with a, a young researcher who was starting his own lab. And I don’t think he had enough budget to hire anybody who could legally drive. So, you know, I ended up, you know, getting this job at the Children’s Hospital and Traumatics programme working for a guy named Zach Kohani. And it was kind of sort of the best possible place to to land, you know, if you were going to end up getting into health. IT sort of this was the mid 1990s. So, you know, the web had just been invented, but it was very new and no one had really thought about how it might apply in healthcare. And Zach had actually started to build really, we think the first web interface to an EMR, the back end EMR was in VMS, you know, and so one of the server assessor software tools. So it was like Oracle 6 on, on, on vax. And, you know, we’re just trying to, it was this cool new thing. And you know, it was still, you know, great, great backgrounds by default and no JavaScript and no, no interactivity.

And but there were a whole lot of people doing really innovative things like, you know, John Halabka, who went on to do all kinds of stuff, was doing his fellowship just across the street at Methodical Deaconess and clinical Informatics. And and so it was this sort of amazing problem space and these really, really smart people who were got me very excited around using technology to create software that helped people. And you know, the joke that I probably have told too much is there’s this great cool idea, but I like, I realised by the time I’m like, done with college, this is all going to be solved and I’m going to have to go find something else. So yeah.

00:05:31 Tony
That.

00:05:32 Will
That was that was my, I was certain I was wrong about that in 1996. And so that was really where I got started. So, you know, very lucky, you know, frankly, for me, I’m very grateful that I had that opportunity. And and Zach, you know, was an incredible mentor and this sort of continued to be, in fact, I think that’s how I ended up at CMS because that was one of the connections years later that led me going down to Washington. But it was sort of through someone I worked at Children’s Hospital that I met my first business partners, a guy named Marty Streeter, and he had started a company called Advantage to help drug companies do a better job placing clinical trials. So again, this was sort of a domain that I knew nothing about. I did not. I, I, I knew that drugs had clinical trials and this like this would have been a big data company if that term had actually existed at the time. And the idea was that we would take data from pharma companies around which doctors did a good job conducting A clinical trial. So recruiting patients and providing good high quality data back and we would make it easier to use those physicians who had access to a rich patient pool.

So we actually started integrating some claims data and this was really not been doing for trial feasibility at that point. And we’d combine it all and we’d send it back out to all the different pharma companies. None of those companies turned out to want to contribute their own data into a pool, but we’d built some pretty cool software, again like early use of the web to both gather data and make it available within the companies. So we sort of ended up building really the first SAS platform sold in the pharma. And again, this was the late 1990s. So and we realised it was it was, it was an important use case, but it was also a way for Merck and a couple of other big players kind of get their feet wet, you know, in the web, both as a way to interact with doctors and then as a way to deliver data internally. So it was sort of a reasonably high value, low risk use case that they could sort of get their heads around and pour it on.

And you know, thinking about that, what it’s such a different company that we built then we ended up selling it to Parxel back in 2002. So, you know, we, you know, we had, I had about a seven-year run with it, but it would have been very, it would have been so much easier to have built that Apple built that platform if we’d had things like the national provider identifier. We, we spent, you know, vast about years and, and lots of money and lots of effort just trying to be able to create a good provider directory. We had data from FDA on who had participated in a clinical trial and then we went out, we found all these other data sources and we enriched it. And, and when the NPI stuff came out a few years later, I was looking at, and I almost went and restarted the company, you know, or started a new version of it because we could have just between that and then NCI had clinicaltrials.gov, which had a lot of other really bad contacts. And suddenly it became a whole different, you know, different ball game because of, of standardisation.

So that was, that was really a career pivot for me. I mean, I, I, when I started on that, I still kind of thought of myself as a software guy and I, I started writing about technology and wrote some books on software development. And, but I was also, you know, still finishing college and I studied healthcare economics with a professor in Howard Foreman. And sort of between that and working with pharma companies, I kind of had a switch flip and became a healthcare person who spent a lot of time doing software. So, you know, I will not spend as much time on the rest of my career. But, but you know, after that, I, you know, I, I, I did, I decided to go to Graduate School. There was this great programme at MIT, which I wish still existed and, and a lot of the alumni are still talking about how we can bring it back called the Biomedical Enterprise Programme. We’re combined an MBA and then with a master’s degree from the Harvard MIT Health Sciences and Technology division.

And I think they created this because they had a bunch of people going to medical school and then going into venture capital and they didn’t want to use up all the Med school slots on the but small cohort. You know, some really innovative people, you know, some of whom have gone on to do some pretty, pretty impressive stuff more broadly in the in the health tech space. And you have become, you know, both on the health IT side and on the life sciences side. And this while I was there, I got to go back to Children’s Hospital. I also took some time off and came down to Washington, got involved with a group called DOSSIA that was building out of patient health record and actually ended up sort of scaling, you know, some delivery operations within the the children’s and Harvard settings so that we could actually build tools that people could actually use it at more scale. So, yeah, everything else kind of followed on from that. You know, by the time you were done with me, I was really focused on just healthcare, healthcare cost curves and how we could make those change. I mean, I, the great opportunity for me about spending time at CMS was like everybody was thinking about healthcare costs and how you could bring, how you could do more with different resources and how you can be creative around what we what we paid for.

We had all these of course, limitations on, you know, what statutorily we were allowed to you to do. And I remember, you know, being a lot of conversations around rural telemedicine and how that happened. I decide from hospital to hospital and finding that I think it’s a 20 something very frustrating that we couldn’t go off and do do more there. And and so everything else kind of came sort of came out of that. I mean, I, I got to, I got very interested in healthcare user experience, which led to helping build a problem oriented medical record company. So you know, this concept really from the 1970s when I got a doctor, Larry Weed, around organising clinical information around a patient’s problem list, you know, around what you’re billing for or just around a narrative node. It feels very simple, but the physicians who had the luxury of doing the record keeping this way, either because they had software that let them do it or because they had jobs that allowed them to spend a lot of time curating data, I loved the way that it impacted their clinical care. So this was actually the only EMR package that David Blumenthal, who was the national coordinator for health IT under under Obama, had ever used. You know he.

00:12:40 Tony
Great. I remember that, yeah.

00:12:42 Will
And so he would get up in front of these big audiences and say, my EHR makes me a better doctor and everyone would laugh at him. But with this software actually was true. I mean, there were a lot of physicians who used it who would talk about how it helped them, you know, surface the right information at the right time and improve their clinical, their care quality. So that was really exciting. And and then I got to go to Fitbit, as you mentioned, and we ended up spending about four years there, which was very different because that was a, a small health outcomes impact. But for 30 million, 40 million people at a time, you know, and then, you know, finally, when I got to the medically home, kind of knitted everything together. It was wearables, it was electronic health records and it was clinical trials and life sciences. And so, you know, really, really rewarding time because we also were trying to do something that was new. So we got to create this whole new, new category of, of health IT software. So that was way too much. So I apologise for taking probably most of our interview on that. I, I, I hope.

00:13:53 Tony
That’s right. Well, it’s been nice talking to you. Unfortunately, we have a lot more to, to talk about. And I think the no, I think this was, it was helpful because what it did is, is show kind of the depth and breadth of what you’ve been doing over the last almost 30 years now, because there’s obviously a lot of ways to look at the healthcare either from a cost care and other types of main actors that are involved in it. So I’m going to, I’m going to turn to one of those areas which you spoke about a moment ago, which is really I, I was going to originally do this question based on the, the Consumer Focus.

And it is, still will be. But really the whole area of home health care as as we look at that and it’s, it’s evolved over the years you’ve been involved in it, obviously from medically home and from your own personal experiences. I know you told me about your dad and some of the things that he’s challenges he faced, But I guess the question that I have for you is we’re kind of in a high tech, high touch mode that people talk about. But most of what I’ve seen has been more the need for high touch rather than than a high tech. So I guess the, the question I want to really ask you is so, So what are the challenges around bringing technology into home healthcare in a way that we’re dealing with people at levels that aren’t exactly sophisticated health IT people. We’re dealing with the home health aides that you know, have a very not, not a high level of tech certainly background they may except for maybe their own personal use of technology. We have patients in a lot of cases that are elderly or certainly not people who have a lot of experience with technology. And then we have a healthcare system that sometimes from a payment standpoint is still not giving payments to some of these areas around technology that that maybe would be helpful. But they haven’t quite gotten to the point where the industry is willing to pay for it or CMS. So it’s kind of a roundabout way of asking you, what do you think of all this?

00:16:24 Will
Oh boy, you thought my last answer was was long. You know, there is a boy, there’s a tonne there. And so I’m going to take a, you know, maybe maybe even a little bit, sort of a little little bit of random. I mean, I so I mean patience first, though, because that’s, that’s always a good place to start. I mean, there’s a lot of variability, you know, in, you know, what a, you know, patient can absorb in terms of TAC, there lots of remote patient management companies that became reimbursable. I’ve talked with a lot of companies that are focused on enabling children to usually adult children to monitor older family members. And, and probably what’s boiled down to is that I’ve, I’ve met 90 year olds who can fully manipulate, you know, all of these tools more effectively than you know most 30 year olds.

00:17:28 Tony
Oh, absolutely. I’ve seen a few myself, yeah.

00:17:31 Will
And I’ve also met, and I think, you know, that’s great. I think it can be a little misleading because you know, of course, there are also lots of people for whom the technology is very intimidating and you know, certainly people who never learned how to learn how to navigate technology. I think that’s really the differentiator for, for a lot of people of any age is like, if you, if you have the opportunity to learn how to explore what a system can do and you can, you know, it’s, you can do eye tracking studies just with people using basic web applications. You don’t, you don’t need to sort of go into a healthcare domain to see this.

Some people are very good at exploring and they’re fearless and they’ll push the button and see what happens. And some people are are, you know, don’t, don’t have that wiring. So I can medically home. We spent a lot of time, you know, on the user experience of the tablet that we put in the home because that was the patient’s conduit to care, you know, and being able to get that patient talking with a, with a nurse in a clinical command centre. You know, our average was 17 seconds from when they hit the button to when they were talking to a nurse. And so we had to address, you know, people with palsies who couldn’t necessarily tap the screen. We couldn’t expect that there would be any infrastructure in the home at all to tap into. So it was, it was an interesting design challenge because we, we ended up building it a very simple tablet experience where we probably a third of the screen to tap here to call your care team. And we put a lot of the smarts on the back end so that we could equip the nurses and the doctors with context about that patient so that they could be that interface so that, you know, you can build something, you know, And I, I think our key persona there was, you know, we was a 90 year old with a fever, you know, you know, with. So you, you get to do no training, you get to do no onboarding. And then we we built another hole mechanism where if that didn’t work, we actually had a telephone handset that they could pick up and be connected to the nursing commencement.

00:19:47 Tony
OK, right.

00:19:48 Will
Right. And then we got into other human factors issues like being able to like finding a thermometer connected thermometer that was accurate enough and easy to use enough was almost impossible. You know, a lot of them were sort of the pistol grip thermometers that were designed for taking someone else’s temperature. Most people can’t bend their risks back far enough to be able to use a device like that on themselves. So, so a lot of human factors, a lot of user design. So, you know, I think that’s a big part of it, you know, is it’s possible and it’s so much easier now. I mean, even like say, an example that I like is lithium ion battery packs. Actually, we wanted to make sure that all this connectivity that we put into the home would still work right when we started treating patients, you couldn’t buy an affordable lightweight portable lithium ion battery pack that you could put into a patient’s home as it as an uninterruptible power, uninterruptible power supply.

Those came out really during COVID and, and so now, now you can’t. And so then there was a whole set of additional capabilities, including things like home infusion that we could more reliably put into the home because we could guarantee that we would have power the way that you would end up, you know, in a hospital setting. So yeah, I think there is a lot of excitement around the technology enablement. That all being said, it’s very much about people. It’s very much around home health aides and, and building. There are a lot of companies that have gone out to sort of build a marketplace for home health aid services or automate the business of hiring a home health aide and you’re paying them on W2. And you know, but there simply aren’t enough home health aides, you know, and there aren’t enough of any other level of profession getting home health aides from place to place outside of extremely dense urban populations. Like a lot of these problems have been solved fairly well in Manhattan. You can, you can get to 5 patients on the subway on the same block, but you go to Los Angeles where my like, for instance, where my, my parents live and you know, it’s a 30 minute, you know, drive to go to, to go 2 miles, right, right. And so for all of these in home services, that includes, you know, in home urgent care and in home phlebotomy, you know, getting the economics right is really difficult. So there are like there are plenty of companies in LA that will come and we’ll do an in home, you know, IV infusion or IV vitamin infusion, you know, for $700.00 a visit because that’s, it’s a luxury good. But you can’t get someone to come give you a flu shot, you know, and because you just can’t, you can’t make the travel time for a nurse work, you know, in that setting. So I think again, a lot of technology opportunity, a lot of logistics opportunities to improve things there.

But you know, probably the biggest thing, you know, I, you and I spent a bunch of time looking at sniffs, you know, back at CMS and, and I, you know, I’ve, I’ve spent a lot of time in them since. And there’s some great skilled nursing facilities out there. Oh, absolutely. And there are some not so great skilled nursing is out there. And, and it’s not always the fault of the human beings when they’re not that great. It’s the, it’s the fault of against some very, very difficult economics. And I do think the technology can help out a lot there. And again, I don’t think it’s necessarily around AII don’t think it’s about, you know, ambient documentation or although I’ve been in sniffs where there isn’t a documentation system. Yeah. And yeah, I think one of the biggest indicators of quality when you go into one of these spaces is do you see people with the screen on the wall and are they in point click care? Are they in one of those tools? If yes, that’s a good sign. You know, a lot of sniffs are still running on paper.

So lots of medication errors, lots of, you know, you know, you know, lots of handoff issues that come, come out of that. And then just again, a staffing at the staffing level, they’re not using technology to make sure that, you know, the, the nurse or the CNA’s are going to the right rooms at the right time. And, you know, I’ve, I’ve asked a bunch of them about this and, you know, often they cite HIPAA and it has nothing to do with HIPAA, you know, and but it does have to do with, you know, they’re, they’re so under resourced and so understaffed that they don’t really have access to people who can help them build a good regulatory and compliance framework. And to understand and explain to their patients that, you know, with the right level of consent, you know, having a system that has a video camera and can monitor a patient in a bed and can identify if a memory care patient who has physical frailties is trying to get out of bed.

There’s nothing in HIPAA that prevents you from doing that. And you can build it extremely, of course, not protecting piece of infrastructure. And there’s lots of hospitals that are doing this that supports that work. And I’m just not seeing it, you know, in the sniff setting. And I don’t know whether some of that is because of the staffing mandates that are in place that those are driving all the budget people, you know, or whether it’s just hasn’t been seen as an attractive market by firms that could make it easy to equip a facility that way. So I think there’s a lot of opportunity and I, I, I do see companies, you know, that are investing fairly heavily into that. But so I think those are probably 3 aspects, you know, to the question. And and we could, I could, I could keep going.

00:26:14 Tony
No, I think you, you unpacked a lot there. I I think in terms of the, the home health area, I think as well as in the, in the skilled nursing facilities and, and, and other, well, let’s say assisted living, other types of environments that people reside in after they can’t stay in their home any longer, whether it’s a nursing home or something that’s less full time care. But the, the problem, I think from the home standpoint, I do think you mentioned the pandemic. I think that did help in terms of telehealth. I think a lot more people tried telehealth. The government relax the lot of the regulations around telehealth. So I think that helped. But I think that the point you, you bring out is part of the problem around home health is the fact if you don’t live in a densely populated area. And of course most of America is not densely populated. I mean we have, we have a lot of areas of dense population, but a lot of America’s is rural or certainly far less populated than some of the major areas like the New York, Chicago’s, LA’s, etcetera. But as you mentioned, even in LA, because it’s much more of a spread out urban area, it doesn’t benefit from a lot of that. OK. So another thing I wanted to talk to you about, well, we sell workflow areas both on the payer and the provider side. You’ve had a lot of experience with that. And of course, one of the big things that people are talking about in terms of impact in that workflow is the recent CMS prior authorization regulation that’s going to make some major changes to that area. And a lot of people are supportive of it. And then others have concerns about it, mainly from the payer side, not from the provider side. So what are your thoughts on major workflow challenges and opportunities and especially with the payers and you know what, what, what is complex, it’s become easy and vice versa that you’ve seen over the years.

00:28:28 Will
So I guess first on, on the prior authorization sort of general utilisation management API requirements, I was really excited, you know, when those came out. I mean, again, just from a, you know, a patient and A and a provider happiness perspective, you know, those are huge opportunities to make some really complicated stuff really straightforward. And, and you know, when you’re working with companies in oncology or in metabolic medicine, you know, where the prior authorization burden is still pretty substantial. And actually my wife is a surgeon. So, you know, even, you know, in that category, there’s still a surprising amount.

Oh, absolutely, you know, and so you’re picking up on that theme. I guess it like how do you build tools that make, you know, address the potential physician work shortage by making it less unpleasant to be a physician? You know, I think prior authorization automation, you know, is hugely important. I also think that, you know, there’s a lot that can be done to, you know, reduce the prior authorization burden overall based on data. Like if we know that 90% of prior auth requests are being accepted in a particular pattern, maybe we don’t need a prior auth that was sort of after all. And so coming back to that idea of, again, how do you build a great user experience? You know, being able to have an API that injects the necessary decision and the necessary follow up right into the point of care. Let’s the people who are designing those clinical experiences, you know, design them in a way that’s that’s much more, you know, physician and patient friendly. So, so you know, you know, big, big, big fan, big fan of those, you know, and, and I think it also means that a lot of companies right now that are trying to do AI for prior authorization automation.

There may not be a whole lot there for them, you know, in the next couple years. A lot of these companies reached out, you know, both because of Spring Tide and some of the other work that I do and you know, they’ve done stuff that’s technically very interesting and very, but it’s not very defensible, you know, and in some cases it may be solving a problem, you know, like, you know, OCR and handwritten clinical notes that just isn’t going to be a problem, you know, five years. So from an autumn. So that being said, from an automation perspective, I’m really excited about automating basically everything else. So the the biggest challenge for smaller digital health companies, especially ones that are trying to go into a provider or market or repair market, you know, is distribution, you know, and what distribution is, how do you, how do you get integrated into a clinician workflow? And you know, Epic is a huge gatekeeper here and some of the other firms are too, you know, Athena, any clinical works and Cerner. But but you know, it’s usually Epic that people are talking about and you know, they’ve, they’ve made some progress and they’ve lost APIs and, and they’ve been compliant with the regulations that have come out, But it’s still very difficult, you know, as a third party to, you know, play with Epic, you and you really need a health system partner who’s going to go in and be an an active sponsor for you in that relationship. And you’re going to be limited.

I mean, if Epic doesn’t have an API or if it’s not an area that may feel needs third party collaboration, there’s not going to be a lot of that. So that’s where some, like some of these agentic AI capabilities are. Yeah, they’re, they’re a little dangerous. I’m a little nervous about them from an accuracy perspective, but they do make it so much easier to automate away the swivel chairs, right, right. Because that is how a lot of these companies go to market is they’ll they’ll build something that sits completely separately from the EHR. You know, we, we did that initially in medically home. We built a platform that solved this complicated problem of coordinating all this, you know, high acuity care being delivered into the home. And then we also solved another problem of how do you make all of these vendor relationships look like an inpatient bill? Because, you know, these were high acuity admissions that were being, you know, billed under AUB four. And they weren’t 1500, they weren’t an Oasis. They were, you know, an inpatient hospital bill.

But they were being provided by skilled nursing companies and home health companies and mobile phlebotomy companies and and medical meals companies that if they were used to sending a bill at all, you knew how to send a 1500. You know, so we had to pull all of that record keeping, all that accounting together. And so we built a swivel, A swivel chair, you know, and then we were able to slowly integrate it into the provider work flows and start taking impatient orders and all that. Even, you know, from five years later, we would have approached building those tools and building that network very differently.

Robotic process automation is not a new technology. No, thanks for doing it 25 years ago, but it is so much easier to create a very flexible RPA implementation today. So when I look at that healthcare provider organisations, you know, I think that’s really automating really sort of boring stuff, but automating it in a way that allows novel clinical models to be experimented with a little bit more easily. Like that’s, that’s what’s exciting. Yeah. So you can take stuff that you know, might have been a nurse operating at the bottom of their licence, you know, automate that away, turn it into a review process rather than a data entry process and you know, let them focus on things that are more clinical, that are more clinically relevant.

00:34:42 Tony
And I’m assuming though with AI that’s going to help even more once that becomes more integrated into the workflow.

00:34:49 Will
I think so. I mean, like the the best example that I’m seeing right now is is you have a chart biopsy a lot of data about, you know, every, every patient that’s interesting enough to need someone’s attention has a lot of data around them around patients who aren’t interesting may not have a lot of data. And so just being able to get some hints to find this sort of to create the directory that lets you say, well, if you’re only going to look at, you know, one note out of the 40, this is the note to look at because the alternative is that you’re not going to look at any of the notes. So from a quality perspective and a quality of care perspective, you’re only making it better. Yeah, because you’re taking a resource that wouldn’t be utilised at all and you’re making it utilisable. So that’s like, that’s so that’s where I get the most excited about it. I don’t, I don’t obviously get too excited about it as a, as a diagnosis tool because you know, it can be right and accurate and 8090% of scenarios and then when it goes off the rails, it goes way off the rails.

00:36:04 Tony
Yeah, well, that’s that’s a life watering problem if it happens with that.

00:36:10 Will
And, you know, just, you know, as an example, I mean, I was looking at, you know, a platform that did clinical, clinical summations and I asked the vendor how they ensure that the LLM that was creating that summary didn’t hallucinate the Med list. Yeah. And, and in this case, I actually think that they did have a way to do it. And, and it’s they had access to the source of truth for, you know, that patient’s clinical medication list. And I I think they were templating it, but the person who was doing the demo didn’t know that, you know, and wasn’t it was able to talk about how old we will, we could swap in clawed, you know, if you don’t like Jackie, but you know that that wasn’t the question, No.

00:36:56 Tony
No, no, no, that’s that’s a little disconcerting.

00:37:01 Will
So, so I think the people talking about applications of AI in these clinical or even semi clinical settings, you know, and suggesting that certain things can be automated, it’s really incumbent on them to understand the limitations of the tool, be able to educate and build confidence, you know, in the user community about about what those what those edges look like.

00:37:29 Tony
Right. And and it’s been part of the problem and I saw this years ago when I was at IBM is people would oversell capabilities and then when it didn’t work the way people thought it would work. And this was we’re of course talking Watson Health back then. Then it just create a, a sea of disillusionment throughout the whole clinical community about its capabilities. Whereas they’re kind of like what you’re saying is you just have to have people that understand what it can and can’t do and work with the end user, whether it’s a provider or provider team or or some some other care setting and you know, go from it that way.

00:38:13 Will
I mean, one of the best recent AI applications that I’ve seen was predictive modelling of medication use in an ICU. You know, it, it allowed them to pre position stuff from the pharmacy on the floor rather than have to run meds up and down from the pharmacy. You know, saved a tonne of nursing time. You know, probably had a positive patient outcomes impact mayor. You know, they were high performing organisations. So, you know, they were probably already taking care of the patients well enough, but saved a bunch of time, saved a bunch of money.

00:38:49 Tony
Right, right. And and that’s exactly the type of outcome we’re looking for in this case, not just something that’s a great technology, but something where you actually have measurements that show exactly where it saved time and money and quality improvements, that type of thing. So I’m going to move on to another area fairly quickly. I know we’re running short on time, but I did want to get some of your thoughts about data. I mean, there’s been a lot of growth and data over the past number of years, types of data, our ability to get to data, our ability to get more longitudinal data, non clinical data as well as clinical data. But we also have a lot of silos. We’ve had a lot of consolidations in the industry pickling the payer side. And I think in some ways we have more data than ever and we have more ways to utilise that data. But there still seems to be a lot of challenges in in our ability to take that data, regardless of where it came from, and be able to utilise it in a way that improves quality and and. Efficiency, I guess for one of a better way of parting of of healthcare. So what are your thoughts on that?

00:40:09 Will
I think the most critical thing you know, for any organisation thinking about healthcare data is to understand what you’re going to use it for. So, and lots of different use cases, you know, understanding patient populations for value based care cohorts, identifying clinical trial subjects, you know, with payers, there’s a, a lot of data that’s around heaters measurements, as well as supporting CM and utilisation management activities. And I see a lot of organisations don’t actually understand their full, their full data ecosystem. So, you know, one of the things to do is we actually work with companies helping them map that out. And we’ll get a whole bunch of people in a conference room and sometimes it’ll take two hours and sometimes it’ll take 3 days. And we’ll actually start working either from right to left or from left to right, depending on sort of where the organisation feels most comfortable mapping out their data flows. You know, and, and everyone is surprised, you know, and like we, we did this with a 15 person company a couple months ago. We’ve done it with some some large healthcare enterprises at how complicated the graph gets and how quickly that graph gets complicated.

00:41:24 Tony
Oh.

00:41:26 Will
Yeah. And then in some cases, you know, you take a line on that graph, you know, lab data in Tahitis data warehouse, and you blow that line up into a whole document, all, all, all its own. But that exercise has been very, very useful to identify hotspots and sort of what we call sort of, you know, impediments to knowledge generation because you know, the data isn’t important to the insights that are that are important. So that’s kind of where we start and kind of where I keep sort of encouraging the industry to start. Then you get into things like clinical informatics, you know, huge amount of work that’s been done on code sets and standards and standard representations over the last 1520 years. ICD 10 is much more illustrative than ICD 9. We’re finally seeing pretty widespread adoption blink, you know, you know, not as widespread as I’d like, but we’re at least getting closer. And, and again, lab systems are, are interesting because they were one of the first things to be automated, which meant that everybody designed their own code system, you know, and you know, I was in one big, one large health system and they had 32 separate ways at which point we stopped counting to order AC reactive protein assay. Yeah. So there they didn’t have a single reporting view, you know, of their utilisation of different volume tests of different volumes.

So yeah, that is one of the places where AI is already starting to help out a lot because it is great for doing 90% of the reconciliation of a lab formula and which then allows you to do reporting in a clear way. There are companies like Atropos, which spun out of Stanford that have invested in building infrastructure to rearrange healthcare data out of the standard data warehouse and into a longitudinal format. So you know what their platform allows you to do. And I don’t have a stake in this, by the way. I just think it’s really neat. And I read the paper in, in Jamia is in close to natural language, say I want to find a cohort of patients who, you know, are over 60 and have taken these three drugs and have this adverse effect, you know, within six months, you know, after that use demographic criteria, which in sort of a traditional health analytics data warehouse is going to be a lot of sequel. It’s going to be very custom.

It’s going to be a lot of analyst time. And you know, you know, with tools like this, which just take advantage of the fact that compute is cheap, you know, so you’re back 20 years ago, we had Oracle 7 databases and they were running on Solaris and we were fine tuning every index. And, and what companies like Snowflake figured out is that, you know, computers cheap and you can move it around and you can build your analytics platforms in a way that maybe use a lot more raw computing cycles, but are much more human friendly, you know, and allow you to be less efficient, allow you to organ, but allow you to organise data in a way that makes it easier to get the answers out that people want. And like we’re seeing things like, you know, version control databases, you know, which are great for master data management, especially, you know, if you’re a vendor, it’s a company called Dolt, which makes a My sequel version that behaves like a git version control system. And if you are managing that different formulary mapping for every single customer that you have as a healthcare SAS company, you can now build your tool to manage those mappings against your own internal vocabulary expensively. I mean, it’s, it’s something, I mean, I, that’s another company I have, no, I have no stake in, but I like, because I’ve wanted that tool for about 15 years, you know, and they finally went and build it, you know, so it was a lot of exciting stuff happening on the data processing level. And I think Snowflake really started a resurgence of this. And then a lot of stuff happening on the health and traumatics and standardisation level.

And then just a lot of realisation within organisations that you need a good chief data officer, you need a governance. You know, when you do these mapping exercises, you discover that, you know, often the problems are technology problems, they’re shift left problems where people upstream, you know, running the production systems, you know, aren’t taking into account what the people downstream using the data need. And the people downstream using the data aren’t doing a good job of communicating their needs and the business justification behind them back to the people who are generating the data at the source. So, so I think there’s a, there’s a a lot of confusion, there’s a lot of volume, but there’s also a lot of good, good work that’s being done both on that sort of base infrastructure side, but also just best practises of how you build these teams.

00:46:59 Tony
Yeah, I think, I think that’s been a real difference. I agree. And if we, we hadn’t gotten that improvement with all the data that’s coming in and from all the places is coming from nowadays, we never would be able to handle what it’s hard enough to handle what it is. But like you said, just just simple data governance, simple practises and processes and really looking at mapping the sources of the data and who utilises that data. Just just doing things like that, what we call, you know, good hygiene, making sure that the, the quality of the data you get, the timeliness and just just some of the things that, that help you. And then you then you can bring in the technology. But if you don’t have some of the basic mapping and, and basic infrastructure in place to support the the use of that data, it’s you’re going to have all kinds of problems and particularly in the industry where there’s a lot of consolidation going on. So if you buy another healthcare company, then you have to deal with the entire issue around redundancy and just bringing together some very potentially disparate platforms that creates all kinds of problems and potential disasters.

00:48:21 Will
Yeah, exactly. And we’ve had, there’s a lot of, there are a lot of roll ups happening, you know, in the hot space right now. And, and in some cases, you know, it’s company companies actually making decisions about who to buy based on the EHR platform. You know, they’re also on Athena. So he’s here, you know, which means your EHR choice. If you’re building a, you know, a tech enabled value based care company, it is a strategic consideration that your investors care about. I mean, like we, one of the things we do at new fires, we work with PE and BC firms and we help them out with due diligence on different projects. And that is very much one of the key things they ask us about is, you know, are they picking the right clinical technology layer? It’s not should they go build an EHR? Should I build any HR is 99% of the time no. But but you know, are they picking the right one?

00:49:22 Tony
Right. Absolutely. Yeah, that’s and, and and that makes sense because of if you don’t then you’re really setting yourself up for, if not failure, certainly a lot of a lot of issues and problems. All right, well, well, we’re going to turn now we’re we’re getting towards the end. So I want to just hitch with a couple quick questions. So when you’re not thinking about the health IT and the the world of healthcare, what do you do when you’re not working on that kind of stuff?

00:49:53 Will
We’ll see. So we have my wife and I. We have three adorable cats and only one of them has an Instagram account. So that, that does keep us a little bit busy. But, you know, big thing is that, you know, I, I grew up around photographers and my, my father was actually the first American allowed back into Vietnam after the war as a, as a journalist. So, so, you know, kind of grew up around people who would go to really interesting places and meet interesting people and, you know, make pictures of them. So, you know, I, I, I would wander off with his cameras. And so I, I feel I don’t do the same kind of work that that he did, but I still, I still do a lot of that. Recently have gotten back into shooting film and going into, because that’s, I got, he taught me when I was a kid, he taught me how to print and how to do everything, everything by hand. And so I have a, a 25, you know, year old film scanner, the last good Nikon scanner that was every night. And there’s a retired optoelectronic engineer in Virginia up by Dulles airport whose hobby is restoring these scanners. So I, I shipped it down to him, He fixed it up for me, shipped it back, got all the old cameras out of the basement and found the one place around here that still processes the film. And you know, I’ve been, I’ve been having a lot of fun.

00:51:17 Tony
Yeah, going. Oh, I bet, Yeah.

00:51:20 Will
And then I, I, I, we travel a lot. So new fire and we have offices in Europe and Latin America and and Asia. So professionally, I get to travel a lot. My wife and I like to like to.

00:51:30 Tony
Oh, that’s nice.

00:51:32 Will
So I tend to, you know, cart cart cameras around on on those trips too. OK, great. I haven’t gotten back to putting film through the X-ray machine yet, but.

00:51:43 Tony
So on the topics we discussed today, well, are there certain go to publications or blogs or podcasts or or other sources of information you go to the keep you up to date on things or where some of the listeners could go to to learn more about some of the topics we talked about today?

00:52:05 Will
Well, I think for, for really sort of understanding, you know, the implications of AI in healthcare. The New England Journal of Medicine launched their AI journal that’s actually edited by Zach Kohane, you know, guy who gave that first job in health IT way, way, way back when. And, and so that I’ve I’ve really enjoyed and they actually have a podcast that goes with it. So, you know, both the journal and the Department of Biomedical Informatics podcast, you know, are sort of great ways to understand how people are thinking about using some of this technology in a clinical setting, but also going a little bit deeper on like, how does this actually work, you know, and, and what are the what are the boundaries of value that these tools can deliver in a real clinical setting? So yeah, I think that’s probably the best place to go to really get get deep on AI and healthcare. On the delivery side. I still, you know, I subscribe to the Politico health policy and digital newsletters back when I was at CMS and I never clicked the unsubscribe button on that. So those are very useful and, and frankly, Mobi health news, you know, which I used to come out of him. So I’m not sure if it still does does a really good job of just sort of keeping up with the day-to-day and the digital health space. And then I, I, I would, I would be committing malpractice if I didn’t mention the, the new fire blog and podcast where so we sort of rotate that around and we bring in interesting people from our advisory universe and, and beyond to just sort of talk about what’s important to them. So we’ve done episodes on clinical informatics. We had Ron or Wongsay, who used to be at CAQH and has now defined health come to talk about provider directories. We have the head of data from Morgan Health, you know, and so we, we try to kind of wrap that back into this sort of sorts of problems and challenges that the customers that we work with are solving. So, so you know, you’ll occasionally get to listen to me, but you don’t have to listen to me all the time because we.

00:54:17 Tony
I’ll definitely have to look at the subscribing to that. So just a couple, we do this kind of quick rapid fire at the end. So 22 quick questions. One of you were starting a, a company in the in the health IT or healthcare space today. What would you do? What would you start?

00:54:37 Will
So I’d look and I’d look and I’d look in two areas. The, you know, first is generational support. You know, we talked about sniff and home health aides. You know, I don’t have kids, you know, and yeah, I’m very lucky and that my mother, you know, is in a position where she can really quarterback my father’s care. But, yeah, 30 years from now, if I need the same kind of quarterback, I’m, you know, you know, I don’t want my wife to have to deal with that, you know, and we don’t have kids who would be able to take that on for us. And there are a lot of other people that are in the same boat. So, yeah, that’s an area that I would be looking at very hard, you know, if I was starting something new is how do you, how do you create support systems and that’s technology and it’s people and it’s insurance and it’s process to, you know, let people age and safety. I think it’s the term I use that right. And then on sort of the more light hearted side, you know, my, my team like kids me a lot about being obsessed with building, you know, medical spas because my wife who is a surgeon, you know, also opened a Med spa, you know, as an adjunct to her sort of core clinical practise. So I got to run the EHR selection process, which terrified all the vendors because they gave them my background and they basically went like, Oh my God, he actually knows, you know, he, he knows what questions to ask. But, and of course, it also put me in a no win situation because I had to pick an EMR for my wife, you know, and the software out there for that domain, It’s all terrible. So we picked 1:00. We, we hate it. I’m not going to shame them because I kind of hope they’ll come to new fire and ask us to fix it. So if I didn’t have a lot of other stuff going on, I think it would be hard to keep myself from building one last EMR.

00:56:27 Tony
Right. So if you had one thing you could change in healthcare today, what would it be? I mean, you kind of went over some of these, but yeah.

00:56:41 Will
If I could truly, if I could really, really think big, I mean, you know, I would want the incentives aligned. Yeah. I, I think probably the, the paper that turned me fully into a healthcare person from a software person was a David Coupler economics paper from probably 2004 or 2005 talking about the impact of disease management, you know, and how most disease management programmes for diabetes didn’t make sense and didn’t succeed because the benefit came after the insurance company’s holding period. You know, on that, on that member. So there’s so much that we can do with technology and with care delivery and that we just don’t get to do because nobody’s allowed to pay for it at the right time or the right way.

00:57:35 Tony
Right.

00:57:35 Will
So, so truly, if I could fix anything, that would be what I would fix. And then if I didn’t get to fix something at that level, I would open up as much access to clinical workflows, clinical tools, clinical APIs as possible. I think there’s a public good argument, you know, that could even be made, and I think that would just unlock a tremendous amount of innovation within the health system.

00:58:03 Tony
Absolutely, absolutely. Well, well, thank you for spending the last hour with us. Appreciate it and best of luck to you.

More from this show

Subscribe

Episode 7