A national registry could help improve transparency

The use of artificial intelligence in hospitals is growing rapidly and with little transparency, making it impossible to track how specific products are affecting the cost and quality of care. Whether AI is monitored is entirely up to individual health systems.

“Every company is doing it differently, and many health systems aren’t doing it,” says Mike Pensina, chief data scientist at Duke Health. The lack of systematic oversight of AI in healthcare poses safety and financial risks, he says, because poorly performing products can easily escape scrutiny.

Pensina is one of several people proposing to fill this void with a new approach: a national registry of AI tools that would list where the products are being used and provide context about their development and performance.This isn’t an entirely new idea.