A new editorial in NEJM AI argues that approving clinical artificial intelligence tools is not enough — regulators need to keep watching how they perform once they’re in use.
The New England Journal of Medicine’s AI publication has run an editorial calling for continuous monitoring of artificial intelligence tools used in clinical settings, saying current regulatory frameworks fall short once a product reaches deployment. The piece argues that oversight shouldn’t stop at the point of approval.
According to the editorial, existing approaches to regulating clinical AI are limited because they focus heavily on pre-approval testing rather than tracking how these tools behave in real-world healthcare environments. The authors argue for what they describe as postdeployment oversight — keeping a close eye on AI tools after they’ve gone live in hospitals and clinics.
That matters for patients. AI tools used in diagnosis, triage, or treatment planning may perform differently across different populations, hospital settings, or over time as clinical practice changes. A system tested in one context doesn’t always behave the same way in another.
The NEJM editorial does not call for slowing down adoption of clinical AI. But it does make the case that regulators need to build continuous review into the system — not treat approval as a finish line.
For NHS services in Kent and across England, where AI tools are increasingly being piloted for tasks ranging from radiology image reading to appointment triage, the question of how those tools are monitored after rollout is a live one. NHS England and the Medicines and Healthcare products Regulatory Agency (MHRA) are the bodies responsible for overseeing medical devices and digital health tools in the UK.
Anyone with concerns about a clinical AI tool used in their care can raise them with their GP, their hospital’s Patient Advice and Liaison Service (PALS), or by contacting NHS 111.
Source: @NEJM