STAT+: U.K. unveils recommendations for regulating AI in medicine — SkimNews
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- A U.K. commission unveiled 44 recommendations on Thursday for regulating AI in medicine, tackling the global regulatory challenge of continuously reviewing adaptive products after they reach the market rather than clearing them once.
- The recommendations seek to balance AI's clinical potential — such as reviewing millions of scans annually to track eye health in diabetes patients — against safety, longitudinal device performance tracking, and equitable patient treatment.
- Several proposals would enable staged authorization of new AI models, creating a pathway for incremental regulatory review as products develop.
- Other recommendations call for building a provider reporting infrastructure so clinics can document when AI tools malfunction or potentially inhibit patient care during real-world use.
- The framework is specifically designed for AI models that learn and adapt over time, performing differently depending on the clinical setting — a category that breaks traditional static-device regulatory assumptions.
Why it matters: With 44 recommendations covering both staged authorization and mandatory post-market reporting, the U.K. is attempting the first concrete regulatory structure for medical AI that changes after deployment — giving health authorities worldwide a potential template for governing adaptive software rather than frozen devices.
Ask SkimNews
