Fixing The Flaws In Clinical Research w/ Faro Health Founder & CEO, Scott Chetham

Clinical trials are the backbone of medical progress, but they've become increasingly complex, costly, and slow. Can technology smooth the process without compromising rigor?

Scott Chetham, Founder and CEO of Faro Health and former clinical operations leader at Verily, explains why outdated processes bog down research and how Faro’s digital trial platform is making studies faster, more efficient, and more patient-friendly. By transforming static documents into connected, intelligent systems, Faro helps sponsors and regulators streamline design, reduce amendments, and improve trial outcomes.


🎙️⚕️ABOUT SCOTT CHETHAM
Committed, strongly clinical and technical research executive with proven experience in developing innovative solutions to complex medical problems.

Experience spanning private startups, hospitals, university research labs, venture capital and patient advocacy groups.

Established strength in forging partnerships between leading academic medical centers, community clinics, advocacy groups and government organizations to solve complex clinical syndromes.

Strong technical skills in large complex data sets, electromagnetics and human disease.

🎙️⚕️ABOUT HEALTH BIZ PODCAST
HealthBiz is a CareTalk podcast that delivers in-depth interviews on healthcare business, technology, and policy with entrepreneurs and CEOs. Host David E. Williams — president of the healthcare strategy consulting boutique Health Business Group — is also a board member, investor in private healthcare companies, and author of the Health Business Blog. Known for his strategic insights and sharp humor, David offers a refreshing break from the usual healthcare industry BS.

GET IN TOUCH
Follow CareTalk on LinkedIn
Become a CareTalk sponsor
Guest appearance requests
Visit us on the web
Subscribe to the CareTalk Newsletter

Transcript

David: 0:03

Clinical trials are the backbone of medical progress, but they've become increasingly complex, costly, and slow. Can technology smooth the process without compromising rigor? Hi everyone. I'm David Williams, president of Strategy consulting firm, health Business Group, and host of the Health Biz Podcast, where I interview top healthcare leaders about their lives and careers. My guest today is Scott Chetham, CEO of Faro Health, and a former clinical operations leader at Google's Verily Scott saw firsthand how inefficient and burdensome today's trials have become, and he decided to do something about it. Do you like this show? If so, please subscribe and leave a review. Scott, welcome to the Health Biz Podcast. Hi. Thanks for having me on. Yeah. I wonder how did you first get involved in healthcare and in clinical research in particular?

Scott: 0:53

Um, it goes back to, uh, when I was in, in university in Australia, I, while I was training, I worked as a coordinator, uh, enrolling patients in clinical trials. Then I actually went on in, um, during my postgraduate study. Actually ended up working in a cath lab training physiologist, and I started enrolling my own patients in clinical trials. Uh, actually have been around long enough that, um, I was the assistant for the, uh, second dual chamber implantable, converted defibrillator in the world by Medtronic. Um, and, and because one of the beauties of being from Brisbane, Australia is. I worked at a tertiary referral center, and we had a captured population, so we had zero attrition. We couldn't lose patients in a clinical trial because you couldn't go anywhere else in the entire state. So obviously, like a lot of the big sponsor companies who do this type of research used to love coming to us, uh, because of our, you know, the engagement from patients. And you just frankly couldn't go anywhere. So we, you, you were, we were a very captured population. Um, and that's kind of how I got into the industry. It was more of the nature of where I was, um, doing my postgraduate study at St. Andrews Ho at St. Andrews Hospital and kind of my placement there and, uh, training. So it was fun. And one day a couple of my bosses pulled me aside, the director of cardiology, and said, Scott. If you do this job, you're gonna be doing the same thing again and again. You like research, you should do that. Um, as a, as a helping hand to guide me and that, that's kind of what steered me in this direction.

David: 2:29

It's always helpful, uh, if you've got someone like that and 'cause it probably would've been easier for him to say, oh, I got this guy. He is like a robot. Let's just keep going and not tell him about the big bad world out there. I mean, I've been involved in some clinical trials in Australia. It's an interesting place for that. 'cause as you said, first of all. I mean, it's a, each state is also pretty large. Uh, they're not going anywhere, and also costs aren't too high. And from a regulatory standpoint, it's kind of easy to get started sometimes, which is good for companies that are just getting their funding and need to get some, you know, get some evidence, uh, in place. I always enjoyed it.

Scott: 3:01

Yeah, it's a, I mean, it tends to even today be the, a place you go for your first in human research. Um, because a regulatory barrier while not lower, is just more efficient.

David: 3:12

Yeah.

Scott: 3:13

Uh, having said that, market access to getting payment is terrible. Yeah. Yeah. So choose your poison

David: 3:20

of what you want. Yeah, fair enough. Exactly. Well, you'll get to your trial, you'll, you know, sort of saying, where can I do it? Like you said, it's not lower standards, it's just more efficient. Um, but then you certainly don't want to be making a living there, uh, long-term, at least in the drug business. So then you were, you were at Verily and I'm wondering what you learned there about clinical trials and, you know, structural issues, uh, with the trials.

Scott: 3:44

Yeah, so I mean, verily supports, um, and at the time I was there, a number of the, you know, very public partnerships with some of the largest sponsor or pharma companies on the planet. And they were offering various like services for decentralization. Like how do we help, um, enroll patients more efficiently? How do we collect data, um, more efficiently, uh, when they're not on site? Um, they had a huge baseline study enrolling 10,000 participants over, um, several years measuring a lot of things, um, mental health. But the problem is, is you still inherit what I would say, just poorly designed protocols. And the reason is is I mean as I reflect the, even the clinical trials for the pandemic, those protocols and what happens to participants, so like what happens day to is in a table in Microsoft. And there's no information about what worked before. Because what we tend to do, or the state of the art before we came about is you open what someone did before, or you look at what someone did before, copy and paste, and then you evolve it and you inherit all the mistakes and all the problems from that. And the biggest sin in some regards is the fact that in we redlined things in Microsoft Word and then we clean it up and we check it into a system for regulations and we promptly forget everything we ever learned. Some ways you think working for someone like Verily is because we supported so many large sponsors, you kind of think, well, maybe they're better than me or they do it a different way. And what I realized, we're all doing it the same way. Um, and so the, the, I was like, well, we have to fix that. And 'cause if we can't fix that, it doesn't matter how much more efficient we get in any single point solution, we kind of have to fix our design and then we've got the opportunity to start doing some what I think some really exciting things coming now, which is the ability to automate all the connections to the point systems and I think, and which will really allow in some ways, we do a lot in my field of busy work. I think it'll really allow us to spend time on the strategic part of our roles versus frankly just creating document after document after document, um, and then taking that information, re-synthesize it, and then handing it to someone else. I mean, I think getting out of that part of the job is gonna be a thing that can really help a lot.

David: 6:08

You partly answered this question I think in your, in your previous reply, but I noticed that, you know, clinical trials have become. Quite cumbersome over time. And I would say even especially so in some areas that are particularly critical, like rare disease or immuno-oncology where the patients have, you know, tremendous need and there's also great opportunities for, for breakthrough. Why I, I mean, first, am I right that trials have to become more cumbersome and, and why would that?

Scott: 6:36

It's a couple of reasons. One is the science has gotten more complicated, and so because the science is more complicated, we have to collect more things. As we've moved to rare disease, we have to now spend more time, um, helping. I would say identify the right responders in a rare disease, which means more testing, more identification. Then because of genomics and pharmacogenomics, we have more testing on there to understand like how is the drug interacting? And so science has driven this explosion in, I would say, the amount of procedures or activities we do on patients. I'm choosing my term there correctly. We do them to people. We don't do them with people, unfortunately. We do these things to people. Complexity or data collection is just directly, uh, is directly proportional to our success rate. It's been studied quite extensively, but, uh, unfortunately the biggest predictor of trial success is actually how much data you're collecting. It's not the science and it's petrifying, and so it's a true effect. We are just doing more stuff, but the, and the reason it's so overwhelming is because we do everything by hand. Is one of the last industries I think out there that, you know, it's ripe to be automated is just because, you know, it we're because of the nature of the industry and the quality bar, we've had to do everything by hand. But now I think technology's finally at a point where that that's pivoting and changing.

David: 8:19

So let me, I'm just gonna ask you this because this may be as a, as a devil's advocate 'cause there's a, there's quite a few companies and that have made, you know, good inroads and good money and things like electronic data capture and, you know, other kind of technologies and people applying supposedly AI to clinical trial design. Um, and having different tools that are used for, you know, clinical trial management, and there's just the whole ecosystem, uh, of that kind of e trial, uh, business. So what is it that, first of all, maybe a two part question. Uh, is it fair to, is it really fair to say that it's sort of, you know, done by hand today? That's the starting point. And then what makes your approach different from others that may be purporting to, uh, bring some technology to bear here?

Scott: 9:03

Okay. No, great, great. Couple of questions. So I'll, I'll refine my statement. We have excellent point solutions, excellent point solutions. We have many, we have them in recruiting. We have electronic data capture systems, we have clinical trial management systems. We have, um, a budget system. We have, uh, payment systems. They're all individual point solutions. So what I say is done by hand is I design a protocol. I have a written protocol, but then I have to take it to between 20 and 60 vendors. I send them a PDF and they have to take that, scrape out what they think I meant to program these individual point solutions. And that's what I mean when it's done by hand. Um, because we are data and designs concepts, just simply don't like other fields where you have APIs and you can just hit a button and you can connect things like for payments. That world doesn't exist yet. It's very much coming and it's coming very, very fast. Um, but that's the stuff that's done by hand and that's where there's a lot of inefficiencies. And also because of this disparate nature of the point solutions, it's not a learning system where we can take all this information back on those implementations and fold it back into the design. What worked, what didn't. Um, and that's kind of what I mean by hand. Like if I'm doing, designing a clinical trial. If I wanna know what it'll cost today, I have to hand it to a colleague who'll then do kind of identify all the things in this trial of the data I'm gonna collect, when I'm gonna collect it. And they'll put together a site budget takes about four days for a place to do it. That's what I mean. I think c So the big like standards agency, uh, for coding said at a recent conference on average, uh, a protocol is transcribed about 50 times. And mostly incorrectly. And so it's a, it's a just a, it's a sad fact, uh, of, of the reality. I think how to, I think it's the second part of the question is like, what, what does fire do differently here? That's right. Um, okay, so you can think of us a couple of ways, but the simplest way is rather than a static PDF representation of a trial. I would say what we are is a digital representation of a trial. And what I mean by that is not just, okay, I have a series of columns with a, with a field in it. What I mean is that we have a deeply connected, uh, AI system of knowledge graphs so that I know if I'm collecting, let's say, a chemistry panel, it's almost in every clinical trial. I know that. There's many types of clinically, many types of chemistry panels, and that causes a whole heap of confusion a lot of the time in, uh, trial and protocols. But we know that that is, let's say it was a chem 20. We know all the analyze, we know all the units of measure, we know the link code to tell your lab vendor how to do it. We know the data packet that you'd need to, to send that. This data would have to be transmitted because we also know the CPT code. We can tell you that what this will cost. Uh, instantly as you start designing a trial. So we also know the SNOMED code. So if you wanted to directly extract data, we also have all the C representation. So you can automatically program an ED, C system instead of spending six weeks doing it. So what's different is while it looks, our system looks like word as you are typing. It's intelligent enough to do the reconciliation and everything under the hood that all these processes are being built out for you that can go later, go on and be automated. Uh, and that's what I mean by digital trial. But what gets really exciting is because you have all of this data, you can use now, AI agents to better informed design and say, okay, your patient visit times throughout the trial look like this. We've actually digitized the entire corpus of public domain information and performance for trials into the system now, so we can tell you, oh, if you have a visit time of this and these activities, you, it will lead to this amendment. And by the way, your enrollment rates look like this. And so it directly feeds back to the teams while they're designing it, designing this thing. What this means for a patient, what it means for a site, who would you need to do it, what it'll cost upfront, so you don't have to wait for all of these things. You're getting all this information, and I think the world of it won't be us who solves this automation problem entirely, but there's a world coming where basically I can hit a button and approach system, be programmed. It's public. We're partnered with Veeva to Auto Program, their EDC system. In a fraction of the time that a hit like today is best practice, four to six weeks. Do it in a day. Um, and that's, there's that possibility. Now across the entire ecosystem, we can't obviously do all of that. Uh, but what we can do is empower other vendors to use our information and provide it to them so that they can reduce the time and improve the quality. Because instead of humans transcribing things, reading it, interpreting it, transcribing it. A machine could frankly do a better job of just directly pushing this was the intent.

David: 14:11

Got it.

Scott: 14:12

Um,

David: 14:12

so, you know, a lot of what you talked about upfront was how you've got these word documents, people do the redlining, and then you don't, you lose out on whatever had been learned along the way. Um, and that is sort of assuming that you've got to trial design and then it. Continues all the way through. Another issue, which I heard you touch on a little bit, 'cause you used the Word amendment, uh, is that once a trial gets started, often there's a amendments and then that is very difficult, uh, with the protocol for whoever is implementing it at the trial sites to actually be able to deal with that. And that is, that is a huge problem, uh, at least in my experience. Uh, do you, do you deal with that issue as well?

Scott: 14:53

Amendments. We deal with a part of it. I mean, I, I'll break it up. Um, this is not my stat, but about 50% of amendments are preventable. We can help on that side of things. 'cause amendments fix problems. The other 50% is just net new stuff you've gotta fix, they fix problems. So I think getting that to zero is frankly never gonna happen. Um, but the avoidable ones, um, we can help with, you know, quite a bit. As I said, we absorbed all the corpus of information and all the reasons for amendments. So, uh, some of the things, uh, coming from us shortly is if you said you had an inclusion criteria that this person had needs a. Some sort of biomarker value of X. What we can tell you is if X in a similar therapeutic area or indication has ever changed in an amendment, we can tell you upfront and link you to any protocol in history that's had that change. And then you'll have to make a decision. 'cause it may or may not be relevant to you. And I think this is the, this is where I think AI in some ways shines. It's that workflow modeling that it can tell you things that you can't easily find yourself. It won't tell you what to do. Just the act of that research would take me formally in a role. It would take me weeks to do some of this. Doesn't mean we'll make the right decision. Yeah, yeah, yeah. It's a different,

David: 16:11

no, but it's good. But I think what it also highlights though, is the role of AI in terms of what it will replace. Like you, if you are a senior. Analyst or clinical, you know, clinical trial practitioner. If you have that information, you can apply your judgment to it A lot of the times. The problem is you don't ha really have that information. It's anecdotal. You remember John over here worked on this other trial where they did this, that, or the other. You remember reading about something or hearing about it, a conference, maybe you did it and this is gives you more systematic approach.

Scott: 16:40

Exactly, and I think that's where, where the technology really is right now, at least in this space, is the ability to surface information from huge amounts of scale that is out there and bring it in. The other thing it can do is with agen ai, it can frankly just automate repetitive things very well. Um, in reviews and things like okay, and stuff people miss, like, um, we call them maybe a playbook that, okay, if you have a data retention standard in what you add a country to a trial in an amendment, but the data retention standard in that country is like 22 days versus five in the US it can point things out to do as soon as you do it. So you can do stuff like that relatively well. I don't think we're at the world yet where. I mean, we give a lot of design suggestions, um, with, you know, literature to back it up, but it's not at a point where you would go, alright, I'm instantly gonna do this. What it recommends, it's a, it, you have to, it'll surface things that are, you may not be aware of. But human judgment's not going away. Humans have to have ultimate say. Here also may not have, AI doesn't always have all the context.

David: 17:47

No, it doesn't. So we've been speaking a lot at kind of the, the systemic level here about, you know, the issue with design and how, you know, all the pieces need to be knit together and so on and so forth. Can you bring it down to the level of maybe a, a specific trial to the extent that you can talk about that? What is, you know, kind of different or accelerated thanks to your involvement? Um, to the extent that you can share something specific that would be, I think, helpful.

Scott: 18:12

Yeah, I mean, I can give a couple of really specific examples. Everything to enterprise down to a small couple startups we work with without mentioning names. So what was really interesting is we had a, a customer with pediatric rare disease. What was really interesting, one of the, the outputs we can give is like, well, it sounds simple, but here's the amount of blood you're gonna have to draw every day and act. Also, here is the visit times because of pks, you know, pharmacokinetic sampling. You have to deliver a drug and in pediatrics from two to like 12 year olds. And what's really came out apparent when they looked at it, 'cause the team hadn't thought about it, is. Patients would, I should say these. Um, and our pediatric patients would've had to sit there or a 2-year-old for three consecutive days, for 14 hours in a hospital, or I should say a trial site for this. And the amount of blood they were drawing was way over the limit, um, to, to do the, uh, to do the PK analysis. They called that up front. And what ended up happening, and I was excited, like I learned all this, that, you know, secondhand is they actually took the analytics to the FDA and had a discussion like, this will, this trial will never be enrolled bluntly. Like it's just not possible. And they had to, because it's an approved drug. Adults, so it has to be studied for PS and the FDA actually granted them a novel design where they ended up randomizing the pediatric patients. So no patient ever had to experience more than one pk. That's how they solved it. And it was a really nice negotiation with the FDA, where they actually saw it and visually and was like, oh, okay, this, we get it. Now you're gonna, we agree you can do something a bit novel here to make sure this actually happens and ends up being, hopefully gets approved. I don't know that drug, that trial is enrolling now, but that would've never been, that would've turned into a mess if I. Enterprise is really cost efficiency, and we have, this is part, I mean, we have a public paper with Merck, um, where the savings is over, you know, a bit over a hundred million, uh, through basically I would say improvement of design from a patient-centric approach. What I love about Merck, they never told anyone, go save money. What they told their teams is, is let's focus on the patient experience. Let's focus on what the data's telling us about how long the visits are, how burdensome it's, um, and the site. What can we do to make this trial more efficient? It turns out when you do that exercise, you end up saving money. 'cause a large, a large chunk of it is, do we really need this today? Do I really need to answer this question at phase two A? Do I really need to answer this at this? So it leads, a lot of this data just leads to a more, I would say, enables better discussions among the team to drive decisions. Uh, and we think it's 'cause I. Like yourself and, uh, you may have been on this side when I've sat on the sponsor side myself and um, on trials is you have this fear that I can't go back and collect this again and without data pulling in the other direction, this is what it means for the patient and the site. You tend to fall on the side of fear and while I just collect everything.

David: 21:41

Yeah. Let me get it.

Scott: 21:42

I might need it. Yeah. 'cause I'm there. Yeah. 'cause I might need it. Yeah. And so I think it, it really just helps teams. These analytics and what I've seen is that teams generally will 90 more than 90% of teams, and we've done this enough now, I'm pretty confident it seems to be this, this about the number. We'll, when they see data will change, there's a small number of teams that I want it and I want it, and I'm gonna collect it and I'm, I'm gonna do it no matter what. That exists, unfortunately. But it's the minority. I think most people in this industry to like help, help people.

David: 22:14

So in a way you've anticipated my next question as well in your answer here, which is really about the alignment with regulators and sponsors on the shift toward digital transformation. And you gave me a great example, uh, of FDA, realizing, hey, they may dictate certain things, but uh, the result is gonna be, you know, no trial that's done. And they're willing to, uh, change that, especially if they don't see a risk and they see another way to do it. And on the sponsor side, at least a more enlightened one, like. Uh, Merck seems to do it for the right reasons, and then it ends up having a, you know, financial return on investment. I can imagine, well, first of all, maybe there's more examples of, of people all pulling in that direction, but what are some of the tensions? Because to go back to something you said before about a lot of busy work, you know, that busy work is some fairly well paid people's job that they're fairly happy with and have been, uh, doing for some time and don't have a huge desire to change. Not everyone's like you and wants to stop doing something boring, to do something crazy.

Scott: 23:13

Yeah, I think, uh, I'll, I'll tackle the FDA one first. I think the, there's definitely a move and we've seen this, the FDA is very much into digitization of protocols so that they can take a digital form and then do something with it. In fact, they're part of, they, I know they have an internal initiative on this, um, tells. They're definitely aligning with, I would say, automation of the analysis on their side. I think what is less transparent on the FDA and I think they could help the industry a lot here, and it's in two ways. One is the FDA is still remains to be silent sometimes if you go to uh, FDA review meetings. If you kind of have standard of care in your protocol, and I'm gonna collect this physical exam and this thing and this thing and this thing and this thing, they'll never tell you if they've seen another sponsor that is collecting less data, uh, and is less like, as much more patient friendly. They'll never say anything. In my experience. They'll kind of just let it go. And I think FDA could go further along clarifying for its safety and standard of care what their real stance is. Because otherwise you can end up with, I'm not gonna say the name. There's one of the most studied, studied drugs in history that still has a chemistry, full chemistry panel measured, uh, in every trial. It is, and because it's so successful, it's often used as a, I would say, as a first line therapy and a second drug has added in, but it now has decade of chemistry panels in hundreds and hundreds of trials. What. Why are we doing this? Like there's a lot of things I think the FDA could do again, and I think they're moving in this direction by the way, is to really, I think, help sponsors along by collecting less relevant information, which would help them not having to spend so much time analyzing huge corpus of information so they could go a little bit further there. I also, this my personal. You probably experienced this 21 CFR part 11. If people know that it's e-signature should be updated just for everyone's reference, it's, it's a very one page, very short piece of regulation that describes the audit trail required for electronically sourced and signed data. It hasn't been updated now, it's more than a decade. I was gonna say, and it does limit, eh, the use of like. ER systems in like for a primary endpoint in a like phase three study, you'd have to be very careful 'cause you need validated EHR systems. So I think one tactical thing that the FDA could do very quickly is. Update some of these regulations for the way they want data to be used more at scale, because that does directly affect a lot of what gets done.

David: 26:11

Um, can we talk about, uh, international harmonization as well? We, we were involved in some related to HIV and other, um, antivirals, and there were sometimes contradictions between FDA and European regulators. You'd go to one, they'd say one thing the other, and we actually, you know, kind of brought 'em together. What's happening with harmonization these days? And is, is that a big factor, uh, in what you see?

Scott: 26:33

It's better, but it's not great. I'll give a real example, and it happened recently. So there's an, uh, the FDA's clarified its stance in oncology. Like they rejected, uh, a drug from a sponsor, I think it was yesterday or the day before because of not enough US patient data, but as you said, for European payers and regulators, there's a difference in some of the pro like ePRO like surveys. Um, that, that are actually approved in each jurisdiction, which means because of that you have to reuse country data between them. I have to collect the same information, two, three different ways for different regulators. And that and some of these EPRs are burdensome.

David: 27:14

Yeah. And ePRO, the patient ePRO is electronic patient reported outcomes.

Scott: 27:18

Yeah. And even some of the ones for clinician reported outcomes. 'cause there there's different acceptance. This is a problem because we are, it's not good for patients, it's not good for um, sponsors because it wastes time and money. It would be nice if our regulators, the EMA and FDA, a couple of the others. Actually just agreed to more standard things because this is indirectly hurting everybody. Um, so I think the re it's nice that the regulations themselves are very much harmonized, but the secondary layer on top is like, what is the data required that's not harmonized very well. And so I think, um. There's some work, there's definitely some work to be done there, I think on the re of, of the regulators to, between each other because I, I mean all this design, so we just see like why there's so many pros and that's the reason. Um, yeah.

David: 28:13

Got it. All right. Well, we could go on. I have other questions, but I'm just gonna ask you a final question that I ask everybody, which is whether you have a, a book that you might recommend for our audience, and then if not, or in addition, if maybe if you have some book we recommend to avoid.

Scott: 28:29

Um, okay. No, it's a, it's a great question and the book I'm gonna recommend is the Jolt Effect, and it's recent. And the reason is, like, I've come into this field from a background in clinical development. Like if you'd asked me 20 years ago, like, would it be the city of a software company? Um, I would've laughed. Uh, but it also means I don't come from a background of sales. And, uh, what I would say like sales development and commercial development. And so I think, you know, the j is much along those lines of like how well, and those of us who don't come from a business background. I think it's helpful. Um. So, yeah, that, that's my take. It's, it's one of the more useful things I've read lately.

David: 29:15

Great. Excellent. Well, that's it for yet another episode of the Health Bizz Podcast. I'm David Williams, president of Health Business Group. My guest today has been Scott Chetham, CEO of Faro Health. If you like the show, I hope you will subscribe and leave a review. Meanwhile, Scott, thanks so much for joining me today.

Scott: 29:33

Thanks for having me.