Hello, I’m Dr. Jeff Kingsley and welcome to another edition of Riding in Cars With Researchers. Today we’re going to interview Mr. Eric Gildenhuys. Eric is a tech guy and has been in technology for decades. Eric’s with a company now named Deep 6 who is looking to disrupt the research industry by looking at different ways of finding patients for clinical trials.
How are you disrupting the research industry?
People associate disruption with technology, and we have technology that can do something different. The question that comes up is, “how can we do things differently to help get lifesaving cures to patients quicker?” Deep Six came from an intelligence community environment, from engineers who were able to use artificial intelligence and other technology components to pinpoint valuable information among tons of data. That’s the engine that allows us to do things differently. We started talking to doctors and we realized we could make things go a lot faster and easier for them by helping researchers find and validate patients to be recruited in clinical trials.
Finding patients can be a painful process involving talking to doctors and asking if any of their patients have certain conditions. In the past, they would say “yes” and those patients would get into the trial. Today, when we ask if they have any patients, they have to thoroughly review medical records. Today’s trials could have 10-30 different criteria requirements or there may be exclusion criteria that would disqualify a patient for a trial.
And it’s getting worse. The average I/E (inclusion/exclusion) is more than 70 and it’s getting worse. The endeavor is increasingly more difficult which greatly increases the need for a disruptive technology.
What’s interesting is we just left a meeting talking to clinical research coordinators and found that 85% of their time is reviewing medical records, and most of that time is spent invalidating them. That’s a lot of time. And when we say disruptive, one of the things that I also look at is, they don’t know that something else exists technology wise. That’s the big difference today, I think. And that’s where you’ve identified what we’re doing at Deep 6 and you discovered that this is something that nobody’s thought of or nobody’s doing, and we are doing it in a different way.
Why is the research world so slow to adopt technology that’s being used elsewhere?
I don’t know if it’s slow, but I think it has to do with a couple of components. No. 1 – there are a lot of regulations. When you have regulations, you do not want to step out of bounds. When I meet people in healthcare, I’ve got to be very literal because you’re taking care of patients lives and if it breaks regulation, people are not willing to risk it. No. 2 – healthcare is not a hierarchal organizational structure in the sense that there is not one boss that says everyone must do it a certain way. The doctors are there to take care of the patients. They’re going to do whatever they think is best for the patients. Thus, trying to deploy a technology means you need a lot of doctors on board. Whereas, in the business world, the CEO says, “this is going to help the company.” He’s gotten feedback from everybody and can propagate it down to the rest of the organization. That enables people to rapidly accept new ways of doing things.
And that’s spot on. It’s hard to get doctors to agree. We don’t agree with each other.
I jokingly said that doctors are very similar to lawyers and, some doctors would kick me out if I said that, but the fact is, both people represent the best interest of the patients and, because of that, they’re very similar. That’s what their focus is, which is great, but it makes for great challenges in other areas.
So is Artificial Intelligence (AI) real?
Yes. The use or application of AI is real. But I think that it is something sexy and hot that people use as terminology. But the analogy I use is AI as the engine of whatever we’re going to use. The question is how do we apply that engine? How do we use that engine for what purpose?
How will Deep 6 change how I do clinical research?
Deep 6 is basically creating an ecosystem whereby our platform gets deployed at the provider’s site and it sits adjacent to whatever healthcare information system you use. From a technical standpoint we are taking a copy of every single type of medical record and we’re applying AI to it and other technologies (such as NLP (Natural Language Processing)) to read that information in such a way that we can annotate or grab all the clinical and medical concepts in the patient’s records. That’s the engine, but what does this mean from a researcher’s standpoint?
We are providing a self-service portal where the providers can find and identify patients for trials based on those 70 protocols in a simple way. There is an easy to use tool that enables organizations to have staff be able to find patients. 50% of trials fail because you can’t find patients. I just heard Novartis came out with a recent study that says 80% of trials fail because they can’t find enough patients. If we now have a tool that says, “don’t even bother starting a trial because this tool confirmed there aren’t enough patients” that’s a great first step.
We just found out at one institution, they determined they were spending a half million dollars a year starting up trials that fail.
11% of research sites enroll zero patients. It’s a massive percentage.
And how many of those research sites spent time activating it all – the administrative costs, the staff to find those patients and the cost of maybe doing advertising or other ways to attract patients. That’s a huge amount of money and that gets fed right back up into the life sciences companies which are charging enormous amounts of money for the drugs. So, if we can reduce cost, it benefits the whole ecosystem.
And the other thing is that it used to be rare disease. And it used to be what was called small molecules for large populations. Aspirin is a small molecule, and everybody should take an aspirin. A small molecule for large populations. And now the new paradigm is biologics. The new paradigm is large molecules for small populations. We have research trials going on today where, if you meet all the inclusion and exclusion criteria that we could possibly check before enrollment, we enroll you. Then we do a genetic test and, if you don’t have the right gene, you can’t proceed because we’re now looking for a very small population. So the harder the trials become, increasingly, you’re looking to only put in a few patients into the trial. So you need a better knife, you need a sharper tool to be able to know what trials we should or shouldn’t be doing and to find every potential patient, or you can’t possibly be successful doing research anymore.
I agree. I think whether it’s genetics or 70 criteria, the key thing that’s going to help the person sifting through the patients is, we aren’t just going to look for patients but, we also validate the medical evidence. Do you want to spend time going through a huge stack of hay to find those needles? Or do you want a tool that’s going to say, “here are the needles with maybe a bit of hay.”
Increasingly we are doing ‘just-in-time trials’ where we complete as much of the startup as we can and then we start looking for patients. With these trials, as soon as we find a patient, the sponsor company completes activation, ships the drug, and we complete everything else on our end (since we found a patient). I’m seeing Deep 6 as something that will dramatically change things in favor of those just-in-time type models. Open a site, don’t ship drug, don’t ship all that stuff, and then Deep 6 is nonstop looking for patients. The second you find a patient, you finish activation.
A good example is where you can say we are only going to activate it when we have X number of patients. Certain doctors are always going to know their patients, but when you think about larger institutions or researchers that have partnerships with various providers, tracking all of that information is not going to be easy. Whereas with this system: the patient comes in, the physician’s notes or the pathology report is uploaded into the system, and the patient matches to the criteria. The next morning, your researcher knows this patient is eligible.
Here is a real-life example: At a university hospital in California, one doctor told me that he thinks I have a great tool, but he doesn’t need it. He basically said he knew all his patients and he can invite them to a trial when they came in. That other doctors know what his trial is, and they know what his requirements are, and they will refer patients that fit. Before I could respond, the other doctors in the room told him they don’t ever think of his trial every time a patient comes in. He was stunned.
Deep 6 is a platform that takes that stress away from those doctors to remember empowers whoever is running the trial. This then comes back to also activating when a certain number of patients are in the pool.
With nearly any multifunction practice, if you just track what physician referred the patient, you’ll always see that there’s an internal champion who is the number one referral source. And then you’ve got physicians in the practice who refer zero. Do they have different patients? No, they have the same patients, they are just not paying as much attention to the criteria for enrollment.
This platform takes care of that and imagine that not just happening at one site but across multiple sites. And you are not just talking to sites, you are talking to Sponsors and CROs about rolling this out at a national or global level across an entire trial.
We talked about the Deep 6 ecosystems, and now imagine that we’ve got an application where the providers and the research sites have the ability to use this platform. Now let’s imagine life sciences. Let’s imagine the Astrazeneca, the Pfizers, and the Novartis’ of the world, and the CRO is where they can type a protocol and they type in a self service protocol and it pings all the different sites. They type in the protocol and it pings all the different sites and it tells you how many patients you have that fit these criteria and it comes back in real-time and gives them that information. This enables the life sciences companies to become much more effective in trial design. You’ve got 50 patients that fit to my protocol. They can share that protocol with you. You don’t have to retype it. You can then review it, then you can work with your providers to basically confirm and validate that those patients are recruitable, which goes back to just-in-time. We reviewed their medical evidence and they are recruitable so now all we have to do is bring them in for prescreening. Does that guarantee those patients are all going to be on a trial? Absolutely not, because there’s some prescreening requirements that they might trip up on, but you’ve now got a higher level of confidence that that trial can be successful and be far more productive.
I had conversations with some life science people, and they are spending enormous amount of money on analytics, on data, on dated information and they’re basically doing statistics only a percentage of it will be successful and that percentage is close to two percent. Now with our platform, they can be far more confident that they have a 50-60% chance of being successful, because it’s real time. It’s based on not just coated but unstructured data. It’s the physician’s notes and the pathology reports because nobody’s able to read through all that information and provide you with confirmation that these patients are eligible based on that type of data.
And the physicians don’t have to do anything different. They can still just be a physician. Your patient care is the same, you document everything in the EMR. You don’t need to start using a different platform. You continue to use the EMR that you’re already using today and then this is going to look at labs, path reports, your notes, and what you wrote in the free text field where you’re documenting.
I think you brought up a really good point that, from a physician standpoint, they don’t have to operationally changing anything. One of the things that some people don’t realize is that 80% of the criteria is in that free text information. It’s in the pathology report where the pathologist has dictated something which has been converted into text about what his findings from the test set, the doctor is typing in and sometimes they type it in their own annotation of certain conditions. This gives you the ability to find patients based on that information. So huge, huge disruption.
It’s going to be game changing. I’m a believer. We just demo-ed the product to some of our investigators, physician’s, office staff, research coordinators, and we saw the looks on people’s faces and they are excited about using it. So, congratulations to you and your team. It really is quite awesome.
Research is getting harder and it’s getting harder for good reasons, but it’s unsustainable. We have to implement new strategies to deal with the fact that research is getting harder because if it just keeps getting harder, eventually you slow down R&D and slowing down R&D is not an acceptable outcome. So, this is a right place, right time.
The analogy I use is manually reviewing chart files the way we’d been doing for 20 years and I think there’s a new generation. There’s got to be a better way have and now that you know, when you’ve got enough pain, people are going to change behaviors.
Thank you, Eric!
Thanks for riding along!
Send me topics you want me to talk about at jeff.kingsley@centricityresearch.com!
Subscribe to our blog HERE