The biggest challenge for the FDA’s 10-year-old Sentinel program for post-market safety monitoring of medical products is dealing with a wide range of partners and data sources, according to CDER Deputy Director Robert Ball.
Increased use of natural language processing and machine learning will help meet this challenge, Ball said in a presentation at the Sentinel Initiative annual public workshop sponsored by the Duke Margolis Center for Health Policy.
The agency uses Sentinel’s Active Post-Market Risk Identification and Analysis (ARIA) sufficiency analyses to identify strategic areas that need further development, and it plans to focus on improving outcome validations, beginning with current systems and data.
“That’s where we’re going to be focusing a lot of our effort,” Ball said. “I think we’ll see that being done in an incremental process first, starting with the data that we currently have and the current systems, working very practically on improving traditional chart review efficiencies, then moving into other technologies such as machine learning and natural language processing.”