Matthew Michela: The second thing we’re doing in this big trend in AI, which I think you’re going to see dramatically more of in 2019 to 2021, is adoption of AI. We spent this first generation of AI creating the new algorithms and the computational tools and outcome. If you look at that entire industry, there’s very little adoption.
The reason there’s little adoption is because it’s not a technology solution. I can be the smartest data scientist in Silicon Valley and create an algorithm that gets approval and works. But what I have to get that software into the workflow where they’re not going to change everything they’re doing for one piece of AI. They need to fit.
Our provider customers are big academic medical centers. They’re looking at Life Image and saying, “You’re the interoperable technology platform that touches all data. You’ve been improving our workflow and have been working with us for five years. Why don’t you be the transport mechanism for AI so you can bring those algorithms in on your pipes. We might want to use five different AI companies but we’re not going to do five different independent AI integrations. We don’t have the resource, cost, time, and money to do that.”
We’ve been going through a phase of consolidating vendors the last 10 years. Growth of AI is going to continue to expand the sophistication of those algorithms. We would argue that in the next two to three years, we’re going to see utilization of AI.
If it doesn’t occur, AI companies will disappear. They just can’t continue to fund that kind of technology development. You’re going to see that start to grow as a trend in a big way. The next thing I would point to would be the life sciences’ role especially around drug and device development. We’re spending such an incredible amount of both time and money in that world to develop. It’s incredibly expensive and long for lots of reasons.
The one I would point is there is not a paucity of trials. There are thousands of trials around the world even just for cancer. What they don’t have is patients. Nobody can find patients that are qualified for trials. We have customers in the clinical trial world that may be searching for 70 patients in order to test their innovation. They will look for those patients for three years.
I was literally talking to somebody this week with exactly this use case. What we’re starting to work with those companies now is, since we have access to this data and we can normalize with this other data, we’re playing in this new developing world. There are lots of companies that are looking at improving patient identification, finding patients so that maybe you can take a seven-year drug trial and shorten it to three. Everybody is better off for that.
The other piece of this clinical trial support where we’re seeing really big industry trends is the growth in what’s called real-world data and real-world evidence. If you think about how the FDA thought about trials a couple of years ago, they looked at retrospective data of patients. In some regards, stale data for patients for datasets that might have been very large.
What we have seen is, you run those trials. You do it on a set of patients. It takes a really long time because by the time you collect the data and figure out what’s going on, you might have had a patient event using your drug. By the time the CRO sends a nurse out to pull the records, you’re looking back and you don’t know what happened. Then what happens is the FDA approves these things. They’re clinical experts.
The physicians out there will use it in the real-world and all of a sudden, they get a bad outcome. All of a sudden, patients get hurt. All of a sudden, a drug gets pulled even though it might have gone through exactly the right statistical process in exactly the same way over the course of a decade. In many regards, it’s because the data that it was based on was biased data and retrospective data.
The FDA has been really clear about this. They’re looking to evolve the way that clinical trials are run – to have access to data much sooner and to require the data during the clinical transaction of treatment itself in real-time. They call that real-world data. It’s not, “Hey, this happened six months ago. Go and retrieve it from the database and I’ll figure it out.”
While it’s happening, you understand it in real life across huge populations and pull it through. A third of the clinical trials are running with real-world data and evidence here. That trend is going to continue to explode.