Optimizing the Potential of Artificial Intelligence in Clinical Trials with Jonathan Rowe of ZS and MCC’s Linda Sullivan
What are best strategies for utilizing the promise of machine learning, deep learning, and artificial intelligence (AI) in clinical trials? That’s the key topic explored during MCC’s Linda Sullivan’s interview with Jonathan Rowe, Principal at ZS, where he leads the R&D, quality, and operations risk management function, and former head of clinical development, quality performance, and risk management at Pfizer. In general, Rowe notes, AI systems today are becoming increasingly useful in the industry and are providing encouraging results in developing higher quality, lower risk, protocols based upon learning from the performance of previous trials. AI is currently being developed to be able to review multiple forms of text-based clinical trial information such as audit reports, protocol deviations and monitoring visit reports. Using AI in this regard allows for the identification of issues and risks across a portfolio. Eventually AI might prove effective in people helping capture better text-based information across the trial enterprise. Well-trained AI may ultimately help determine whether particular study events were truly significant or not as well as producing improvements in report writing and enhancing risk management. It’s important to train AI systems to be more sensitive to particular text writing in reports, and we shouldn’t lose that capability, asserts Rowe. “We’re getting pretty good in categorizing findings,” Rowe says. “But we’re not at 100 percent accuracy”, he adds. Sullivan and Rowe also discuss the pros and cons of using AI to assist in looking for trends in CAPA issues and root cause data.