Ahead on the Science, Behind on the Tech: Clinical Trials Wrestle with AI’s Emergence
Earlier this year, Swiss-based pharma giant Novartis began an ambitious transition that company leaders say will ultimately change it from a drug maker to a data management company.
With Nerve Live, a set of artificial intelligence platforms, Novartis has pushed itself to the leading edge of AI technology in clinical trials — incorporating machine learning and advanced data analytics not just into the company’s business units, but into its core operations, too.
“We certainly realized that on the patient side and on the operational side we were sitting on a huge wealth of data and we were under-utilizing the data,” says Luca Finelli, head of predictive analytics and design for global drug development for Novartis. “We believe that through this approach we could build new insights and hopefully use it to predict performance of certain business projects.”
Nerve Live is broken into different “modules,” each for a different area of drug development. One of the key modules is what Novartis calls its “Trial Footprint Optimizer,” a proprietary AI system that’s designed to help Novartis to fast site analysis for its global trials.
Finelli and his staff have spent the past nearly three years feeding data from hundreds of trials involving thousands of sites from around the world. With the Footprint Optimizer, the company can now type in a range of data — the kind of disease or drug they’d like to study, say, or the countries they believe may have promising research — and their computer will spit out recommendations for sites likely to meet enrollment needs or that offer quality records or info researchers may need.
When the company ran simulations as it prepared the Optimizer for launch, it estimated it could improve its patient enrollment by 15 percent.
The Optimizer also brings back near-real time data on trials that are already underway and flags potential problems — like a site isn’t hitting early recruiting goals or has quality issues.
Finelli declined to discuss how much Novartis has budgeted for Nerve Live but acknowledges he has devoted a lot of time and effort to it. (The project has been in the works since 2015.) The biggest obstacle has been to make sure that the data collected and processed is “clean” so the machine can read it clearly.
That’s no easy task given that that Novartis has amassed mountains of records during its decades of business that were mostly written for individual trials not predictive analytics and, so, weren’t always uniform.
Finelli says it’s still too early to say whether Nerve Live will deliver the results the company hopes for — it will take another year or so to gather the kind of data Novartis needs to measure outcomes — but early signs are promising. When Finelli’s team first began deploying the Footprint Optimizer, for instance, about 200 employees were using it regularly. That has ballooned to 2,000 — about one-fifth of the company’s drug development staff.
As slow as that progress may seem, Novartis is light years ahead of the clinical trial industry on this front.
“The industry has nowhere even scratched the surface at adopting AI,” says April Mulroney, WCG Clinical’s chief data officer. “We are so far ahead on advancement in science — we’re doing some really cool stuff, scientifically, in clinical trials — but we’re so far behind on using advanced technologies.”
The Tufts Center for the Study of Drug Development estimates that, all in, it costs around $2.6 billion to bring a new drug to market. AI offers the chance to dramatically trim and cut any waste out of the process, Mulroney says.
So why are most in the industry so slow to embrace it?
“We need to change the way we view data,” Mulroney says, noting that most executives see data as raw information, when in fact it’s a valuable asset.
That’s roughly the bet that Novartis is making.
Earlier this year, the company appointed Bertrand Bodsen — a former executive who changed Argos from a traditional catalogue company into the UK’s third-largest online retailer — as its first-ever chief digital officer. He reports directly to Novartis CEO Vasant Narasimhan.
For Finelli, it’s a sign of Novartis’ commitment to reinventing itself.
“Our ambition,” he says, “is to bring Big Data to every line item in drug development.”