A gap in the evidence
AI tools are entering hospitals at a fast pace. But a new analysis found a gap in the evidence behind them. Researchers looked at 1,357 AI-enabled medical devices cleared by the FDA. Only three had been tested for their effect on patient outcomes. That means almost none had proof they actually improve health. The analysis traced the path from approval to practice. The evidence thins out quickly along the way.
Approval versus real-world results
The FDA clears most AI devices based on technical performance. A device must show it works as intended in a lab setting. That is different from showing it helps patients. Few devices go through randomized trials that measure outcomes. Those trials are expensive and take years. Companies often skip them after clearance. The result is a market full of tools with limited proof. Doctors must decide how to use them with little guidance.
Calls for stronger standards
Researchers said the findings should push regulators and hospitals to ask for more. They want evidence requirements that continue after approval. Hospitals should track how AI tools affect patients in real time. Some health systems have started their own evaluation programs. The FDA has said it is working on new frameworks for AI oversight. Experts said the goal is not to block AI but to make sure it helps. Patients deserve tools that are proven, not just approved.