Artificial intelligence draws a lot of attention for its role in drug discovery, where it’s intended to speed up the process of identifying targets and the molecules that can drug them. But that’s just one of the places where AI is gaining ground in the life sciences. A panel at the MedCity News INVEST Digital Health conference in discussed how AI can solve other pain points for biopharmaceutical companies.
Steve Prewitt, senior vice president and global head of digital innovation at Sumitomo Pharma Americas, said most of the new technologies for clinical trials are for project management. He doesn’t see many good tools that help with the clinical trial strategy—how to design the study to make the tradeoff to improve recruitment and improve outcomes. As an example, he pointed to a Sumitomo study testing a schizophrenia drug in adolescents. The trial required an overnight stay. But Prewitt said it was difficult to get parents of an adolescent with newly diagnosed schizophrenia to commit to an overnight stay. Consequently, study recruitment was difficult.
Prewitt said that in a Phase 3 study for a common indication, most of the cost is not per patient recruitment. The main cost is elapsed time. Every day a trial is running, it’s spending money. Sumitomo does a lot of work trying to shorten trial timelines. For example, the company looks for doctors who might have access to certain patient populations. The firm also does analysis on patient recruitment to find ways to recruit patients faster, which in turn reduces the cost of a study.
The technology of Massive Bio employs AI to match cancer patients to clinical trials. CEO and co-founder Selin Kurnaz said that for a cancer clinical trial testing a drug that does not require a specific biomarker, it costs about $65,000 to find a patient. But for a biomarker-based study, finding each patient costs about $150,000. Kurnaz said she’s seen pharmaceutical companies pay $2 million per patient in studies that require a particular rare biomarker.
“That’s the extent of the cost structure that we’re talking about the burden on pharma to find the right patient in oncology,” she said.
Kurnaz said it takes about 25 minutes to manually prescreen a single patient for a clinical trial. With its technology, Massive Bio is trying to reduce that time to a little over a minute. But Kurnaz noted that even before processing clinical trial participants, the first step is finding them. The company’s technology can mine de-identified patient data to find potential clinical trial participants.
The artificial intelligence platform of Sorcero provides life sciences companies with analysis and insights to inform decision-making in a range of areas, such as regulatory affairs and market access. CEO Dipanwita Das likened the approach to the way the retail industry analyzes data to get insights about customers and customer behavior. One key difference between the retail industry and the life sciences sector is that life sciences data are not housed in any one location. Data can be found in many places, such as electronic health records, payer information, peer reviewed articles, and regulatory bodies.
Despite the data differences, Das said the life sciences industry can still learn from the retail sector. Retailers have reached a level of understanding about the customer preferences, down to the colors that they like for shoes and the channels that they choose to make their purchases. That’s a level of granularity that providers of life sciences services and products need to achieve.
“When you look at that, you see a lot of opportunities, not just [for] AI but technology itself,” Das said.
Photo by MedCity News