Solving clinical research challenges: an IT perspective

The industry has such a need for technology and we just can't go fast enough. It's not just about money - you can sometimes only drive technology development so fast. I even think if we doubled what we're investing, there's still a limit to how fast we can go. I wish we could be out there doing more and faster. I think that's a huge challenge. You meet with customers that have such incredible visions on how they want to bring their medicines to patients and if we could just do more, and faster, I would feel much better.

I think data is our other big challenge. We've been talking a lot about our focus on the capability of ingesting our own data and other data and really providing those scientific insights to our customers. Everyone knows there's so much good in the data that can help drive clinical decisions faster. We have to manage that data to provide those insights. Everybody is trying to do it as fast as they can. Where do you find the right, the richest data - that's another big focus for us and challenge for everybody right now.

MEDS, our data platform, is a great example. That's the Medidata Enterprise Data Store. MEDS is powering the entire platform now. You have two sides to it, you have the side that we're building, that powers everything we're doing. Our other investment is allowing us to bring in other people's data - so if you're one of our customers and you've purchased a data source and you'd like that to be brought together with your other clinical data for us to help you drive a new trial or a scientific decision or some more safety information. We're really investing in that ingestion, that curation process and in providing top quality visualisations to help you drive your business differently.

There's AI in two different areas; from a Medidata perspective I'm going to focus on one. We have a data sciences organisation that focuses on building those AI capabilities and I'm working with my group to think about AI capabilities that we would request from the data sciences group.

Looking at user interfaces of the past, we spent a lot of time a few years ago thinking about how do you get UI being built into your development lifecycle, and not thinking about it as an afterthought. I'm thinking about AI and machine learning in the same way now. You want to design that into powering our platform and we have examples of where we've done that really well with our coder, where we're recommending the codes - we're at 96.7% of our recommended coding. Our data sciences team came up with that algorithm and then gave it to us to bring forward to our customers.

So I'm trying to encourage the teams to think about AI in that way, and think about what the data sciences group can go after next for us to really power the value of our platform. I don't want it to be thought of as this thing on the side, because I think the value comes from this incredible unified platform that we're building and all these instances where we can just bring the power faster and more intelligently with AI capabilities.

If I think about the big problems, even with what we know about AI today, what we're still struggling with is finding the right patients. Can we think about feasibility very differently, today? Absolutely. But I think that's not just identifying the patient in the region in the site, but getting down to that specific patient and making sure AI is a tighter partner with precision medicine.

Today we're still at the very high level - let's get to the right country and the right facility. And we know that facility can recruit really well and so we're going to use that historical information to trust them again, versus really getting to that partnership with precision medicine to get to the right patient. I think that will be a really big problem that AI can help us fix. It will be an evolution of where we are today.

I think we'll start seeing movement on that in probably a year or 18 months, but before it's embedded across the industry it will probably be about three years.

There's still apprehension to moving away from the boots on the ground - and I would say it's more of a Big Pharma kind of view. When you think about the patient, you've got incredibly knowledgeable people of their regions and their locations that work in the pharma companies. And they're not CRAs anymore but they're relationship managers and they know those really important investigators in their regions, and they know how they recruit and how they perform.

They're still very apprehensive of using data because their successes have really been based on their knowledge in their countries. So even when the data is more available and of a higher quality, we have these very knowledgeable people that are still apprehensive. And that's a Big Pharma issue. In mid-size and in CROs they don't have the breadth of people, so they need the services more than Big Pharma which is used to this boots on the ground model.

I experienced that at Lilly. No matter what data we put in front of them, they’ll say, but I know the doctor and I know exactly what study coordinator is going to deliver - I don't want to see a report because I know my relationship is why these people recruit.

I hate to say it, but I think the overall cost of development is driving them to have to change, because they have to reduce that - not only the time it takes to bring a drug to market but also the cost. And so things like that relationship part that's really important to many people, those are the things that they have to take away and they have to replace that with data that drives them to different processes. It's hard though, because it's about replacing people and the things they've done for a long time.

I think so and I hope so. What we're trying to do with real world data and synthetic control is amazing. I look at that not just from Medidata's ability to produce something like that, which I think is phenomenal. I'm really excited that we're working in this space and really motivated to help our customers to change their business with that.

But I'm also very encouraged to see that decisions like that will allow fewer patients to take a placebo. If you think about bettering the lives of patients and the ethical side of what synthetic control can do, it may mean fewer people wind up on a placebo someday. That means we can challenge standards of care faster where there is a better treatment or better quality of life being made available to patients. I'm very excited to be able to invest in that at Medidata and to do that, but I'm even more excited for patients.

They are very open-minded to it. They need help and they need partners with making sure that the regulators will accept it.

And the FDA has been brilliant. They absolutely want to start looking at this and talking about it from the ethical side of helping patients with unmet needs. I think they are going to be most flexible at the beginning where there's a totally unmet need - if you think about compassionate use programmes, if you think about other places where the regulators have given us the ability to move something faster is when it's going to be a novel indication. Oncology of course they're very flexible about.We’re also trying with diabetes; I don’t know if they're going to start there and let us introduce synthetic control as well as they will when there's an unmet need.

Regulators are not a challenge, they're an opportunity for us to work differently. Nobody has said no yet, everybody wants to work on it as an industry.