Medical AI faces gap between diagnostic accuracy and patient outcomes

A growing gap has emerged between the diagnostic capabilities of medical artificial intelligence and proven improvements in patient health outcomes. While AI systems can influence clinical decisions and achieve high diagnostic accuracy in studies, evidence that they lead to better patient recovery remains insufficient. This “proof problem” is drawing scrutiny from regulators like the FDA, which has released a discussion paper regarding risk assessment and postmarket monitoring for AI-enabled medical devices.

In Kenya, a study published in Nature Medicine involving 16 Penda Health facilities found that AI support did not lower treatment failure rates among clinical officers.

Parallel to these clinical challenges, public perception of AI in healthcare is shifting. A survey by the Edelman Trust Institute and the Yale School of Public Health across 13 countries found that 49% of adults believe an AI-savvy layperson could perform at least one medical task as well as or better than a trained professional. This belief is most prevalent among adults aged 18 to 34.