Lokavant Wants to Help Good Drugs Succeed in Clinical Trials
Harry’s guest this week is Rohit Nambisan, CEO of Lokavant, a company that helps drug developers get a better picture of how their clinical trials are progressing. He explains the need for the company’s services with an interesting analogy: these days, Nambisan points out, you can use an app like GrubHub to order a pizza for $20 or $25, and the app will give you a real-time, minute by minute accounting of where the pizza is and when it’s going to arrive at your door. But f you’re a pharmaceutical company running a clinical trial for a new drug, you can spend anywhere from $3 million to $300 million—and still have absolutely no idea when the trial will finish or whether your drug will turn out to be effective. Because there’s little infrastructure for analyzing clinical trial data in midstream or spotting trouble before it arrives, some studies continue long after they should have been canceled, and positive data sometimes gets thrown out because of minor procedural flaws; in the end, 20 to 30 percent of the money drug makers spend on clinical trials goes down the drain, Nambisan says. Lokavant’s platform allows drug makers and clinical research organizations to harmonize the results coming in from study sites, compare it to data from other trials, and discover important signals in the data before it’s too late. For example, a company might discover that it’s not enrolling patients fast enough to complete a trial on schedule, or that the researchers administering the study aren’t following the exact protocols laid out in advance. Such headaches might sound abstract and remote, but poor data management slows down the whole drug development process, which means fewer beneficial new drugs make it to market ever year; that’s the ultimate problem Lokavant is trying to fix.
Please rate and review The Harry Glorikian Show on Apple Podcasts! Here’s how to do that from an iPhone, iPad, or iPod touch:
1. Open the Podcasts app on your iPhone, iPad, or Mac.
2. Navigate to The Harry Glorikian Show podcast. You can find it by searching for it or selecting it from your library. Just note that you’ll have to go to the series page which shows all the episodes, not just the page for a single episode.
3. Scroll down to find the subhead titled “Ratings & Reviews.”
4. Under one of the highlighted reviews, select “Write a Review.”
5. Next, select a star rating at the top — you have the option of choosing between one and five stars.
6. Using the text box at the top, write a title for your review. Then, in the lower text box, write your review. Your review can be up to 300 words long.
7. Once you’ve finished, select “Send” or “Save” in the top-right corner.
8. If you’ve never left a podcast review before, enter a nickname. Your nickname will be displayed next to any reviews you leave from here on out.
9. After selecting a nickname, tap OK. Your review may not be immediately visible.
That’s it! Thanks so much.
Harry Glorikian: Hello. I’m Harry Glorikian, and this is The Harry Glorikian Show, where we explore how technology is changing everything we know about healthcare.
My guest Rohit Nimbasan comes from the worlds of biotech and data science.
And during our interview he made an interesting point.
These days you can use an app like GrubHub to order a pizza for twenty or twenty-five bucks, and the app will give you a real-time, minute by minute accounting of where the pizza is and when it’s going to arrive at your door.
But Nimbasan points out that if you’re a pharmaceutical company running a clinical trial for a new drug, you can spend anywhere from $3 million to $300 million and still have absolutely no idea when the trial will finish or whether your drug will turn out to be effective.
The problem is, there’s just no infrastructure for analyzing clinical trial data in midstream or spotting trouble before it arrives.
As a result, according to Nimbasan, twenty to thirty percent of the money drug makers spend on clinical trials goes down the drain, because of studies that continue long after they should have been canceled, or good data that gets thrown out because of some minor procedural flaw.
Nimbasan is the CEO of a company called Lokavant that wants to change all that.
The company is building a data platform that allows drug makers and clinical research organizations to harmonize the results coming in from study sites, compare it to data from other trials, and discover important signals in the data before it’s too late.
For example, a company might discover that it’s not enrolling patients fast enough to complete a trial on schedule, or that the researchers administering the study aren’t following the exact protocols laid out in advance.
All of those problems can increase the cost of a trial.
They can even lead regulators to deny approval for a drug that might have proved effective if it had been property tested.
For an average healthcare consumer, these kinds of headaches might sound abstract and remote, like something only clinical trial managers would ever have to worry about.
But the fact is poor data management slows down the whole drug development process, which means fewer beneficial new drugs make it to market ever year.
So I think we should all be cheering companies like Lokavant who are trying to fix the process.
Here’s my full interview with Rohit.
Harry Glorikian: Rohit, welcome to the show.
Rohit Nambisan: Thanks, Harry, for having me.
Harry Glorikian: You know, you and I sort of talk off and on all the time about the space and what’s going on, but, you know, having it on the show, I have to step back and sort of forget everything I know about the company and start from scratch. So, you know, can you explain to people Lokavant’s business in a way that would make sense to someone, say, outside of the pharmaceutical industry. In other words, you know, what are the big problems you’re solving for organizations that, say, are running a clinical trial, and how are you solving them?
Rohit Nambisan: Sure, I can do that. I think it bears noting that we should probably step back a little bit and talk about the industry as a whole and where it’s been going, and then I can clarify where Lokavant comes in. So I think as many folks know and for those who don’t, I’ll fill in the blanks. I know you know this area, but in the last, I’d say 15 to 20 years, we’ve been moving in pharmaceutical development away from blockbuster medications, things like diabetes type 2. Right, developing therapies for that and getting each drug developer trying to find a smaller piece of market and larger pie to specialized, niche therapeutic indications. Right. So the way I could probably better started with the diabetes example is it’s no longer diabetes type 2. It’s let’s develop the compounds or therapies for diabetes type 2 patients that are comorbid with that have also chronic kidney disease and are metformin naive, meaning they haven’t taken a particular therapy known as metformin. Right. So it’s a more complex filter criteria, so to speak. Right. And so what happens when the industry moves in that that direction is that when you get into these very niche therapeutic areas, you need to collect particular niche, commensurately niche types of data to validate your hypothesis whether or not this therapy is safe and efficacious through clinical trials. Right.
Rohit Nambisan: And in doing that, you now increased the complexity of the trial greatly, not only in terms of the different types of data collecting, but the amount of different types of data you’re collecting. So now each trial becomes a lot more specialized. Not just specialized therapeutics, but each trial becomes more specialized. Right? And so for that reason, we’ve seen a big challenge as we as we moved across that space. And actually, it’s been really beneficial for patients because now we’re going after, as an industry, we’re going after really niche unmet clinical needs that previously there were no therapies for or being developed for. So it’s a really good thing for a patient perspective, but from the perspective of development, it makes it that much harder. Not only is there a smaller market opportunity, there’s less patients to treat, right, but the complexity, the actual costs of the trial and the complexity of trial has gotten exponentially that much greater. So what Lokavant came out of was we were actually a, shall we say, an internal initiative within Roivant Sciences, which is a company that launches a number of different biotechnology companies and tech companies as well. But better known for biotechnology companies. And we saw a great need to be able to develop therapies for niche indications much faster, much more efficient, much more cost effectively, and also meet the complexities of that trial better through novel data and tech.
Rohit Nambisan: And so what Lokavant is essentially, is a data platform that allows drug developers, pharma, therapy developers, to be able to choose which data sources they need, data types they need for a trial. And we can ingest any of those data sources, we can analyze any of those data sources in a holistic manner and expose patterns or signals that could be beneficial or detrimental to the study on an ongoing basis. And when I say ongoing basis, I mean you’re not waiting until the end of the study. And I guess the best way I can align this is just like my kids do sometimes. You’re not waiting until the last day before your term paper is due, before the project’s due to finish your work, you’re actually assessing, doing bits of it along the way to assess where there may be challenges, which gives you, really, the time to correct issues to manage your trial better. And frankly, each one of these trials now, there are between, what, $2 million and $300 million we’re investing in these single trials at this point. So it’s egregious to me that we do not have the toolset to be able to even identify, pull in that data effectively on an ongoing basis to detect these signals so we can plan effectively to do something about it.
Harry Glorikian: Anybody who’s done a clinical trial knows that there’s a lot of risk. Right. So, you know, can you talk about some of the types of risks you’re trying to help make sure drug developers diminish, for the most part.
Rohit Nambisan: Yeah. So I think the way we start with that is always at the highest level, time, cost and quality, right? So when we talk about time, it’s really important to understand that you’re going to be able to achieve less. For example, I’ll give you a few instances. Target participant accrual, right? Obviously for you to run a trial effectively, you need to have particular types of participants or patients, if it’s a sick population. In a vaccine population, they weren’t necessarily sick. So that’s why I use participants as the term. But you need to make sure that you have path to randomized screening and randomizing these patients for your trial in a given time period. Right. And if that’s if your enrollment is is not on track for the countries and the sites you’ve decided to actually activate the study in, you could, your timeline for your study could be exceptionally extended. Right. So that’s that’s one type of one example of a thing we look at to understand how the timeline looking for the study. Another area on timeline for example and similarly is discontinuation. So you can you could enroll patients fine. But if you’ve high volumes of discontinuation of participants in your study, then what ends up happening is you actually don’t have as many evaluable subjects in your study of some evaluable participants. So you have to recruit or enroll more subjects, right? So that could extend the timeline as well. One aspect of the timeline that really affects the overall market opportunity is oftentimes these therapies are only in under patient for a certain amount of time. So the faster you can get them to market, the faster you can get recoup your return on investment. But also on the patient side, the faster we can get these therapies out through the market to address unmet clinical needs. That’s just one flavor.
Rohit Nambisan: Then we have subsequent types of flavors. When we talk about data quality, making sure the data is actually collected in the way that you stated you wanted it to be collected in the plan and the protocol at the outset of the study, as well as cost implications. Right. So we look at cost implications as well, which is how, what will this, what will the extension of enrollment or bad data quality data do to the overall budget that you had planned for this study? But then when you drill down on the level further, you can get into risk categories, is something we look at quite a bit when we look at things like protocol, adherence, when you’re when you’re collecting this data, as I mentioned, it has to be done per a very prescriptive method that is specified a priori before starting the trial in a protocol. And if it’s not collected in that manner, it can be discounted. So we are tracking the risk to protocol deviations and understanding trends and not only understanding trends within that study, but we’re looking at similar types of studies in this particular therapy area, neurology or say, psychiatry or gastrointestinal type studies and saying, what has been the protocol adherence in studies like yours? And therefore, can we glean some insights about how you are doing in your study based on your comparators in the study as well? But that’s just a small flavor. I could probably wax on for quite some time on this question.
Harry Glorikian: Well, that that brings us to the question — I mean, everything you just said, it brings to the question like, from what I know, the company sort of predicts how clinical trials will go by comparing it against a proprietary data set of, I think I was reading, 2000 past trials, right? So I guess the question becomes, so you’re comparing one to the past of things that are similar, but you know, for everybody who’s listening sort of, you know, where does that data come from in one sense, is it truly proprietary? I mean, that’s what I’m you know, that’s my set of questions at the moment.
Rohit Nambisan: Sure. So I worked for a while, before coming to the life sciences, in the R&D space and the life sciences commercial space. And I think that data sets, are there are proprietary datasets in that space? Very much so. But there is a third party market for that data a little bit more. So then we find life sciences data. It’s really hard to get access to R&D data and as you can imagine, that makes a lot of sense, right? If you’re a drug developer or a pharmaceutical developer that successfully completed a trial, you never want to share that data. Thereafter, you spent millions of dollars investing in the study, if you want. If there are potentially unknown issues that you haven’t identified, would you want to put that at risk? If you are similarly, if you are a therapeutics developer that didn’t meet your endpoints, do you want that information to get out and maybe potentially things that issues that that you should have should not overlook, right, getting out in public, etc.? There’s just a lot of business risk. There’s also IP risk, right? There’s a number of different risks associated with getting that data out. So it’s been not a very straightforward journey to aggregate data in life sciences, R&D. That being said, I think how we approached this was we’ve developed models that are both used for benchmarking, as I mentioned before, comparing against similar trials for particular performance KPIs, so to speak, as well as predictive model generation and machine learning models that require a fair amount of data to train on to actually deliver value.
Rohit Nambisan: And in that model, we’ve talked to a lot of our partners or let’s say folks that leads them before their partners. And we talk to them. We say we have a growing dataset. There’s precedent for this because we’ve done this with other partners, number one. Number two, we’ve worked with them to leverage their data combined with our data, write their enterprise data with our data, because it’s a common, it’s not just one entity’s data that’s going to provide that value. Your performance, your processes, the way you run trials is inherent in your data. And if we don’t leverage that data to train our model to retrain some parts of our models against, we’re not providing you the most value we could be with our predictive models or benchmarking. So with that approach, we’ve been able to do comparative analysis of our data set versus other people’s datasets and then anonymize their data upon having a partnership with them to grow our data assets in a very risk-tolerant manner. Right. All the information about CROs or sponsors or other entities, people running trials is removed from the data and we only leverage that data for the purposes of analytics or generating a benchmark. So none of that data is ever shared. So through that process, over the last, I’d say two years, maybe a little two years and change since we started, we’ve been able to continuously grow this asset and provide greater and greater value with our descriptive diagnostic predictive analytics as well as our benchmarking.
Harry Glorikian: How much money, if you had to guess just to give people like an idea, how much money do you think gets poured down the drain preventably every year, and you could save all this money if you just ran smarter, if you did smarter clinical trial management, if I had to frame it that way.
Rohit Nambisan: Oh, at least I would say we’ve done some back calculations on this and happy to digress into the details of them if warranted, but at least somewhere between 20 to 30 percent of the trial costs right now and depending on the phase and depending on the therapeutic area, again, that could be anywhere from 20 to 30 percent of $3 million to $300 million per study.
Harry Glorikian: Yeah. I mean it’s you know, that’s got to be, I don’t know how many billions that is. I can’t I don’t know exactly how much is being spent annually off the top of my head.
Rohit Nambisan: We believe we’ve done some back of the envelope calculations to show that it is in the billions for sure. Across the across the global pharmaceutical market, we’re looking just just the value proposition and the signal detection we’re bringing to bear is somewhere around $18 to $20 billion, in terms of market opportunity.
Harry Glorikian: I mean, how would you guys run or help a team run a clinical trial in practice? Can you sort of give me a real-world example, maybe de-identified, where you helped the client avoid or mitigate some kind of risk, whether it has to do with patient enrollment or site compliance or safety issues during a trial, any one of those will do.
Rohit Nambisan: Sure. So I think one example that I can bring to bear is working with a large CRO. And with this large CRO, they had a sizable data asset that was not harmonized, so to speak. It was still living in the transactional exports from the source data systems or CSPs. Et cetera. All around. So it was they had a bunch of different hypotheses about where they were proficient, where they were deficient, but nothing validated. So we spent some time with them trying to understand what all their data assets looked like. And we started collecting these different representations of former trials and ongoing trials, and we collected them and we harmonized them. In fact, as I mentioned before, one of our major differentiators is this is creation of a single source of truth. And we take that upon ourselves, too. It’s not like a service, it’s part of our offering, right? Our platform offering. And so what we did was we brought that data together and we it was about, I think 400 to 500 studies worth of data at that point. We harmonized it into what we call our local and canonical data format, which is a single representation for multiple different domains of data, scientific data, operational data, enrollment data, etc. And then we compared that against similar studies in our repository, our growing repository, and said, okay, we can tell you comparatively that you are deficient in these particular areas and you’re very proficient at the various–for example, in this case they were very proficient in achieving first patient in on the timeline that they expected to actually, scratch that, that they were very they were very proficient in actually accruing the subjects by last patient in in the time they were expected to write so they could hit their accrual when they wanted to.
Rohit Nambisan: But when we looked deeper into the data and looked at across like first patient in, the 50 percent enrollment mark for the study and then last patient in for the study, we were able to identify that there was actually a slowdown and a major overcorrection to make up for that. So they were actually hitting what they needed to hit. But as we all probably know, at least in the clinical research phase and any or any budgeting process, being over your budgeting process is bad. Being under your budgeting process is bad, right? So in this case, it’s again the same. They were burning resource and cash and resources to rapidly overcorrect for for a milestone they were not hitting reliably earlier in their studies. And so we realized in that accrual situation we said, okay, what you need is, we’ve identified an error, you’re potentially deficient. What you need is an enrollment forecasting application that brings in the data in real time from your study. Right. And it also combines historical data from our repository in your historical data to seed some prior knowledge about the study. So and it’s automated, fully automated. So every day you can understand where you are in relation to where you need to be. Right? And it’s not a naive straight line kind of curve. It’s basically it’s based on looking at thousands of historical studies in this space and understanding what the curvature of the actual model should look like.
Rohit Nambisan: So we generated that and we were able to actually, in the proof of concept, and this is just one particular example of an application we’ve been able to generate from our clinical trial intelligence platform, we generated that and we were able to, on a study, predict two years out within one month when they would actually really hit the accrual and it was within one month accurate. Now while that was valuable in terms of understanding at the end state, what really the value was of this closed loop model, so to speak, right, is that it is closed loop. It allows them in silica to say, what happens if I open some sites here? What happens if I close some sites? So what happens if I close this country here? How will that affect my plan before I put that into action in the real world, which oftentimes is very, very, first of all, it’s very risky. But second of all, it can yield a number of unknown consequences if you don’t try it before in silico. So I think the approach here was we were able to not only predict these things better and also predict the impact of change orders on the study, that might actually affect the timeline of the study. But we were able to actually provide them an application, an interface by which they could test it all their hypotheses in a virtualized manner before they implemented them. And we’re growing like crazy with that, with that partner right now at that point.
Harry Glorikian: Yeah. And I mean, I mean, you know, in some ways it sounds like, you know, I didn’t get it done and I’m pulling all nighters, like at some point so that I can get it done. Right. So there’s a whole staffing model. And how do you bring this to the attention of everybody so that they don’t drop the ball? Right, because there’s a million other things that might be coming at them at that moment.
Rohit Nambisan: That’s exactly right. Actually, one thing I’ll add to that, given you mentioned the staffing model around it, is that we were born within small biotech. Right. And small biotech is very resource-constrained in its ability to manage and oversee a study. That’s fairly well known. So our approach has always been what I’d like to call machine-assisted human intelligence. We have experts that are human experts that know the space, but they need to be augmented. They need to be able to look at more complex streams of information and have a machine pick out particular salient insights, salient information, and provide that to them so they can process it, reducing degrees of freedom for them to process it.
Harry Glorikian: So just I mean, there are a lot of statistical tools out there now that that for managing risks in clinical trials. So how is the approach that you guys are taking either different or better or both.
Rohit Nambisan: It’s a good question. One way we’ve been able to address this question is that statistical approaches generally require certain amounts of data points to be collected before you can warrant using statistical parameters or assumptions, etc. And so there’s two things at play here. On top of that, I just mentioned, we’re moving into more specialized therapeutic areas, right? So patients per study are smaller, right. And on top of that, when you’re starting out a study which is usually the riskiest points in the study, when you’re early in the study to mid-stage in a study, you cross them with the fact that you have less patients and there are more niche studies, it’s hard to find those patients. Now, your early phase, your riskiest phase, is going to be extended as compared to when you were developing against blockbuster indications. So for a long time in the study, you can’t really reliably use statistical parameters to identify an outlier or identify something as aberrant. And then you need to focus on so the way we’ve done it, we’ve done it in a slightly different way. There’s two approaches. One is we’ve actually developed a pretty complex risk score system that’s based on a set of very different metrics. Think of it as like an array of different KPIs, right? Those KPIs will affect risk differently depending on the type of study you’re in. And they’ll have different weights to those risks of time, cost and quality depending on the study you’re in. So we look at the given study, we’re going to deploy and we say, okay, what are the features that characterize the study? Let’s look in our historical repository against those same features, pull similar, we call look alike studies and we’ll understand how to set those weightings to say protocol deviations at this point in the study are going to impact the overall quality of time. That’s much more for this type of study. So we can basically, for lack of a better term, I guess the simplistic way of saying is we can augment the data that we have coming in from a study, which is small at the outset of the study, with lookalike data to increase the power. Right? So that’s another way to look at this. So we can actually, we have much better power to be able to detect these issues earlier on and reliably confer that to clinical operators and clinical developers who can do something about it.
Harry Glorikian: It would be nice if you had enough data at some point to almost run the whole trial in silico, in a sense. But I think we need a lot more data get there. But, just for everybody that’s listening, sort of as a philosophical point, the reason we put drugs through clinical trials in humans is we simply don’t know whether they’ll work or what the unexpected side effect they might have once you start them on a much larger population. So in that sense, it’s expected, even normal for some drugs, maybe even a lot of drugs, to fail at some point in phase one, phase two or phase three. And as an investor, you know, you don’t want it to fail in phase three. You want it to fail early. So is Lokavant’s goal to reduce the failures or simply help drug developers get to yes or no faster, safer, more cheaply?
Rohit Nambisan: Yeah. So our approach has been initially get yes or no faster, safer, more cheaply, more efficiently, right. As part of that process and actually related to some of the work we have done in the last few months on monitoring scientific risk. Right. You have to be careful about these efficacy analyses because they can unblind the study, especially when you have single or double blind blinded studies. So you have to be careful about this point. But in some circumstances we can actually leverage our analysis on blinded endpoint analysis and understand how particular endpoints are collaborating or not collaborating or trending, to understand if there is any effect whatsoever that’s being generated in the study. So this is early days for us. But to your to your point about the first use case, we are starting to think about that as an opportunity as well, because we found a way to effectively blind the information and still assess the information content to understand if there is any form of efficacy signal being produced. So I think that that is a really valuable way for us to approach the market in the near future. I think the other point here is that if you are cleaning the data, if you are identifying those data quality issues on a more real time basis, you should be able to reduce the time to do an interim analysis. Right. We should be able to — you mentioned fail fast. Right. Failing fast requires you to also assess the data, to understand if there’s an efficacy signal, there’s a safety issue. And if we have these long cycle times before we can actually do an interim analysis. And much of the data indicates that those long cycle times are due to not knowing where the issues are and finding those issues then cleansing them. If we can do that faster, we should be able to do interim analysis much more frequently. Therefore, being able to generate a fail fast scenario.
Harry Glorikian: You could almost, you should be able to set up the system to almost be running it and sort of move the bar on where it is on, “Looks successful,” or “It’s moving down towards failure.” There’s got to be some sort of almost real-time indicator as data is coming in to. You just don’t want humans to jump the gun on that. The interesting thing is, I was looking at one of the blogs you have and you sort of say that one of the main reasons clinical trials are so costly and inefficient is bad data management and a lack of interoperability across data repositories. And, you know, it’s funny because anybody who listens to this show knows that just comes up over. And it doesn’t matter who you are in health care. It is a recurrent theme that for some reason people are not willing to step up and solve. I mean, it has to be a party like yours that comes in and helps clean it up from the outside as opposed to it being cleaned from the inside the way that you would ideally like it to be.
Harry Glorikian: Let’s pause the conversation for a minute to talk about one small but important thing you can do, to help keep the podcast going. And that’s leave a rating and a review for the show on Apple Podcasts.
All you have to do is open the Apple Podcasts app on your smartphone, search for The Harry Glorikian Show, and scroll down to the Ratings & Reviews section. Tap the stars to rate the show, and then tap the link that says Write a Review to leave your comments.
It’ll only take a minute, but you’ll be doing a lot to help other listeners discover the show.
And one more thing. If you like the interviews we do here on the show I know you’ll like my new book, The Future You: How Artificial Intelligence Can Help You Get Healthier, Stress Less, and Live Longer.
It’s a friendly and accessible tour of all the ways today’s information technologies are helping us diagnose diseases faster, treat them more precisely, and create personalized diet and exercise programs to prevent them in the first place.
The book is now available in print and ebook formats. Just go to Amazon or Barnes & Noble and search for The Future You by Harry Glorikian.
And now, back to the show.
Harry Glorikian: So on this show we talk about, you know, how does analytics play into this? So, how do—and I’ve got to start finding new words—but AI and ML come into this picture. What types of tools in the AI toolbox is Lokavant using? What special powers does AI give you to extract predictions from your data set that other people don’t?
Rohit Nambisan: Yeah, I think I think the first piece is, and it’s going to sound interesting in relation to what folks usually talk about in terms of AI and ML, but it’s a harmonized data model, right? When I was working as a data scientist a number of years back, nobody told me all the work that you have to do with data governance and data harmonization. And then when you think about fast forward today where a lot of the actual models themselves are function calls, right? You realize that a lot of the work is actually making sure that data is ready to be analyzed for this particular use case. Right. So it’s not to say that we don’t do a number of different, try different approaches to gradient boosted descent or support vector machines or neural nets, which is actually my background in terms of grad school and research. But we spend a lot of time thinking through how we need to harmonize, create validated data pipelines to harmonize data for use. In this case. And even in that case, a lot of the work we do is a kind of intelligence or artificial intelligence. So when we’re harmonizing the data, we’re looking for views on leveraging multivariate clustering algorithms to actually figure out which particular types of data attributes should be mapped to one particular field.
Rohit Nambisan: So it’s not to say that the data harmonization is devoid of intelligent approaches, it is full of intelligent approaches, but it is an absolute necessity to have the integrity of the data that you need to run those sophisticated front end models, which we run a ton of. But I just I want to call attention to the fact that that is a core asset for Lokavant from the get-go, that Lokavant’s canonical data model and the processes we use to harmonize data to get it into that state has been a core focus because if you can do that—and that is the same model you’re providing to your data science and analytics teams, your product development teams—then you really have that flywheel that you can generate a number of different analyses. For example, I just mentioned that predictive enrollment forecast model that comes off of in our our Lokavant canonical data model. That is something that is a predictive model, leveraging historical data and ongoing study data in an automated model that indexes towards the historical data early in the trial, indexes towards prediction indexes towards ongoing study data as it comes in. And we have more confidence that input over the trial, that’s like an emergent benefit of having the harmonized data harmonize.
Harry Glorikian: So, you know, one has to ask in the age of the coronavirus, right, how has the business of running clinical trials changed since the pandemic? I mean. And how do you guys…is that an advantage or disadvantage? I’m trying to, you know, place where you guys are in the whole realm of how things have hopefully changed for the better.
Rohit Nambisan: Yeah, it’s been quite a tailwind for us actually. And I would say that, number one, it’s been it’s been beneficial to us because there’s just been a lot more scrutiny and interest in clinical research. Not to say there wasn’t before, especially for niche therapeutic areas, but and the fact that we were able to develop and get novel COVID vaccines out pretty rapidly. But there was also a lot of challenges along the way in getting to that point. And also delays and trials and challenges in therapeutics development to address COVID as well. So there’s just been a lot of scrutiny in the last 24 to 30 months on how efficient and how fast and how effective clinical research can be. So just that alone has been beneficial. Now let’s take the next step there and say that all associated with the pandemic, there’s been a great impact to clinical trials across the board, not just COVID trials or therapeutic trials. Patients, participants couldn’t get to sites for site data collection, right. Site staff couldn’t get in there, too, for data entry or site management or site oversight activities. Right. So in general, it’s been a huge boon to those technology groups that have developed, decentralized or direct-to-patient data capture methodologies, thereby lowering the patient burden and the site burden for clinical trials to continue in a pandemic fueled environment. What’s interesting about that as well, when we think about ourselves as both a data type agnostic platform for clinical research as well as an analytics engine, a platform on top of that, you see this huge movement to another type of data, another data, for example, decentralized trial data as another data source.
Rohit Nambisan: And what we’ve seen also is that while there’s been a shift to a lot of decentralized trial collection on most studies, at least 90 percent of studies and above, they’re hybrid, they’re not fully decentralized. So you have to have some site data collection and you have some decentralized data collection. And that makes sense for those things that may make the most sense to lower patient and site burden to administer. Let the patient let the participant be at home. For those that require like biopsies, etc., that require a participant oftentimes to come into the site, let that be the site. The challenge there is now you have these two different complex data streams that are not necessarily harmonized and aggregated. So this is, again, I think that’s been an area where we’ve been able to come in and say we’ll just as a matter of course, you’re doing business, this is another data set to us. We need to bring these two in and we have to also enable comparative analysis against decentralized and traditional site based data collection, because otherwise you’re going to miss insights. You’re going to miss information that are critical to your study.
Harry Glorikian: Yeah, a part of me was just thinking, you know, you guys should buy somebody, like Unlearn AI and go at it together where you can have, you know virtualized patients that you can put into the trial, but that’s… we won’t go there. So let’s step back for though, for a second. So let’s talk about the company’s origin story. Lokavant is one of many companies launch by Roivant, as you mentioned earlier. A Lot of the companies end up with the word “vant.” So can you explain so that people understand: What is Roivant, how it operates, what are vants and and why was Lokavant born. And how did you become president and CEO?
Rohit Nambisan: Sure. So Roivant started about seven years ago. And I should mention Roivant is our parent company. We were founded out of Roivant and spun out as a technology company itself. So Roivant initially started as a company that launched “vants” — nimble, entrepreneurial biotech companies and now health tech companies as well. When I joined Roivant three and a half years ago, Roivant had about 15 different biotech companies. And what was really interesting about their approach is it was therapy agnostic, so it was not that there was a strategic focus or oncology or strategic focus on immunology. There was a focus around identifying compounds that may have been deprioritized in larger pharma companies, which says pharma companies that had a lot of potential and had could address critically unmet clinical needs. And so Roivant would in-license those therapies and start a therapy therapeutically oriented vant. So at the time Axavant it was the new neurological oriented, neurological disease oriented vant. Myovant was the human reproductive oriented, disease oriented vant. Et Cetera. And so now when you think about somebody like myself who comes from the tech world and life sciences, health care technology world, brought into Roivant three and a half years ago, the premise behind Roivant at the time was we can more efficiently develop these therapeutics and have more favorable outcomes leveraging innovative ways of addressing human talent as well as technology. And that latter piece is where obviously I came in and we were starting to look at in my team what are some of the most significant challenges and frequent challenges amongst the vants themselves in running these clinical trials? And then does that map against some of the more significant frequent challenges we see outside in the market? And not surprisingly, there were quite a few particular areas of resonance.
Rohit Nambisan: At that point in time, they’re about 54, 45 programs being run by Roivant. And so it was across a variety of therapeutic areas. And I guess the thing that hit us in the face primarily was I guess the best way I could say it is you can order a pizza, right? You can understand what is it, a $25 investment, $20 investment. Maybe it’s gone up since then, since I ordered a pizza. But the point is that you can understand what time it was ordered, when it was when they said they were going to deliver it to you, and you can track it. And most of these apps now [show it] along its destination to a chain of custody to get to you. We were we could spend $3 to $50 million on any given trial and we were at struggling with our partners to actually identify what is the current state of enrollment in the last week? What is the current state of discontinuation? Where are we at with our with these particular sites in this region? Why are we seeing high screen failure rates, etc.? Right. That’s egregious to me. That’s just that should not be the case.
Rohit Nambisan: We are fairly frustrated with that. And then even when we when even at Roivant or even in my former experiences at Novartis or other pharma, when we brought in a source system to say, okay, well, we’re going to have a representation of data ourselves, right? So that we can understand what’s going on. Invariably what happened is you would have one source system here and then a duplicate version of that sort of system at the CRO or another vendor that’s working with you. You spent your entire time trying to figure out which was the source of truth, because they’re spending all your time doing data reconciliation, saying, is that really accurate? Is that really the signal? So that didn’t work either. So we felt pretty frustrated about this. We initially tried not to build it ourselves. We worked with a few different collaborators outside of Roivant and tech vendors, etc., and we were fairly frustrated with what we came back with there. So at that point we started thinking, if we can’t buy it, we need to take a lead user innovation approach to address this. So we started out with the data platform, as I mentioned to you, and we built that capability to connect, ingest and map from any source, deliver that within a canonical data model, one single canonical data model. And then initially we did a bunch of bespoke analysis on top of that for a few different biotech vants.
Rohit Nambisan: That went really well. Some of the external collaborators looked to Roivant at that point we said we’d like to work with this technology outside of the Roivant family, and we realized we were on to something, and we externally launched the company in January of 2020, which was very interesting time and year to launch a company. That being said, we spent the first, I’d say, just under two years, really focused on externally subsidized R&D phase. We’re pretty fortunate to have some partners that invested in us in that phase, and we focused on first one particular application in response and we talked a lot about risk. But then we also realized that the needs across different companies, different vendors, etc. for managing clinical trials are very varied. So we realized what we need to really build as generalized on that first application we built and create a highly configurable analytics platform on top of this data platform so that we could actually analyze many different things and configure it for use for any particular customer. And so now we built across, I’d say seven or six or seven different use cases now, and we’ve deployed most of them and we’re continuing to aggregate information in a true product sense where the biggest pain points in the market and how do we build or configure a version of the platform and the platform to address that. And at the same time, we’re delivering on global trials with a number of pharma studies as well as on the side of the vendors working through them to deploy on studies as well.
Harry Glorikian: So in a perfect world, right, if you had access to all the relevant data, if every drug developer in the world was taking advantage of your services, how would it change the business of clinical trials? What would the outcomes look like? Would it be like you get more drugs approved every year, at a lower cost, fewer disaster failures, I mean. What changes for the industry and for patients?
Rohit Nambisan: Yeah. I think the first piece is you would reduce—and this is a lofty question so I’m going to answer with a lofty response—the first thing to note is that, and we touched on this earlier, I think you’d see fewer bigger failures in the analytics phase. You’d be able to identify earlier on, both in terms of the lifecycle of a compound, right? So from phase one to phase three or even phase four, but especially within the study itself, you’d be able to identify that there would be an issue in the study earlier on and you could kill it early on. So that’s one one aspect I think would be that’s important to note. The other thing I think you would identify is less operational issues. So I think one in six trials across the globe failed just because of operational issues. And when I mean operational issues, I mean the protocol and the plans at the outset of a study say need to administer the study following these steps. And when those steps are not followed, there’s compliance risk. And therefore, when there’s enough compliance rates to throw out the data or you have to not submit the study.
Rohit Nambisan: And so one in six is, it’s not that small. And so if we’re tracking, if we’re more rigorously tracking both what is happening and what could happen, right, based on the indication, leading indicators of risk across time, cost and quality, we should basically never see — that’s a that’s one of our major goals — never see a trial fail just because of an operational reason. Not to mention, how can you go to the patients with unmet clinical needs in a particular indication in particular disease and say, “Oh, I’m sorry, while the drug probably was effective, we just couldn’t get it out into the market this time. And it’s going to take us another trial, potentially.” A lot of times folks don’t actually resurrect the failed study, a failed therapy. So even if they resurrected it and said it was because of an operational issue, “Oh, you’ve got to wait another six years.” That’s just not acceptable. So I think those are the two components that come top of mind. And I think early in our in our tenure, our mission was no trial should fail due to operational error.
Harry Glorikian: What is the path to financial success for a company like Lokavant? Is it to just keep growing? To go public? To get acquired by a maybe by a big pharma. What’s the path?
Rohit Nambisan: It’s a good question. I think folks that that know exactly what their exit strategy are probably, for lack of a better term, often deluded. But I will say that we’ve seen a lot of growth. Not only during, there’s been a lot of interest in Lokavant during the pandemic, I mentioned we were in this externally subsidized R&D phase, we were actively trying not to do too much externally. We wanted to figure out how to set up the platform for success. Coming out of that phase, in the last six months, we’ve seen an incredible amount of traction externally. And so I think we are still in the path of doing it on a growth trajectory ourselves. What does that mean in terms of opportunities to collaborate both commercially and partner and strategically? Well, it means that we can only do as much as we can, even if we continue to grow. There’s data out in the market and partners that have access to that data that we would love to collaborate with. If that means that we need to be more strategic in our approach to what Lokavant can do or how to structure Lokavant, we’ll do that just because we need to actually achieve our mission, which is to have no trials fail due to operate operational error. Right. And so I think that requires more data. That requires more data science. We have a lean, very, very proficient data science team. So I think there will be opportunities for strategic collaboration, but it’s all related to the mission of bringing this clinical trial intelligence platform to address operational and other risks in a study as effectively as possible.
Harry Glorikian: You know, one of the things that crosses my mind is you could also use this from an investing perspective to analyze a trial that’s going through its paces against historical information and determine, give it a weighting of probability of success versus failure from an investment perspective, that that seems attractive to me.
Rohit Nambisan: Yeah. So that’s an interesting point to bring up. There are folks now asking us in the market about what we’ve been informed firmly in trial execution stage. Folks are asking us to move into feasibility and effectively feasibility. Is that the planning of the study? Tell me with this particular configuration of sites, countries and for this indication, knowing the standard of care in different countries, knowing the approach to clinical care, not just clinical research, how successful would this study be? Right. And obviously, the success of a study, when you think about biotech, the success of a study is the success of the company. When you think when you go up the market, depending on the study, it can still have incredible impacts, the success of the company. So there is definitely an afferent towards the investing world and financial. I think at first we probably take a progressive step towards that by moving into trial planning analytics in this manner and then validating our approach against progress in space and seeing how we can continue to grow in that sector.
Harry Glorikian: Well, Rohit, it was great having you on the show. I hope everybody enjoyed our discussion. You know, a lot of problems to solve in this industry. So there’s there’s no lack of opportunity from, you know, businesses that need to get started and the data that needs to be optimized to help move the process forward. But, you know, luckily, everybody I talk to on the show, that’s the direction we’re all moving. So hopefully we’ll get there faster, because I’m not getting any younger. So, so good drugs are going to be needed at some point. So good to have you here. And I can’t wish you and the team at Lokavant, you know, more success.
Rohit Nambisan: Thanks, Harry, for having me on the show. It was wonderful to be here.
Harry Glorikian: That’s it for this week’s episode.
You can find a full transcript of this episode as well as the full archive of episodes of The Harry Glorikian Show and MoneyBall Medicine at our website. Just go to glorikian.com and click on the tab Podcasts.
I’d like to thank our listeners for boosting The Harry Glorikian Show into the top three percent of global podcasts.
If you want to be sure to get every new episode of the show automatically, be sure to open Apple Podcasts or your favorite podcast player and hit follow or subscribe.
Don’t forget to leave us a rating and review on Apple Podcasts.
And we always love to hear from listeners on Twitter, where you can find me at hglorikian.
Thanks for listening, stay healthy, and be sure to tune in two weeks from now for our next interview.