Clinicians Rarely Monitor Clinical AI After Launch

A Flinders University review found no studies reported formal processes for clinicians' ongoing involvement with healthcare AI systems.

Updated on Nov. 20, 2026 in Artificial Intelligence

Clinicians Rarely Monitor Clinical AI After Launch

Live Poll

Would you feel comfortable with an AI tool helping decide your medical care?

A review by Flinders University found clinicians were commonly involved in developing and checking clinical AI systems but were rarely involved in their ongoing monitoring. None of the reviewed studies reported formal processes for continued clinician involvement.

Why it matters

The 'human in the loop' concept remains poorly defined in healthcare AI, making it difficult to establish what oversight involves. The review's authors propose clearer reporting so that the role of human reviewers across an AI system's lifecycle can be understood.

Eleven studies described clinicians reviewing AI systems, while only two reported clinician involvement in maintenance. Many reviewed studies relied on small datasets and input from a handful of clinicians, leaving the scale of real-world oversight unclear.

The players

Flinders University

An Australian public university based in Adelaide, South Australia, which conducted the review of clinical AI studies.

The details

The review, published in the International Journal of Medical Informatics, found that activities such as labelling information and designing system rules commonly involved clinicians. It proposes a reporting checklist asking researchers to specify who contributes to an AI tool, what they do, why their input is necessary, and how it affects the system, along with documenting limitations.

Timeline

  1. Article on the review findings published on 20 November 2026.

Big Picture

The review highlights a gap between the 'human in the loop' principle that is widely cited in clinical AI and the lack of documented processes for clinicians to keep overseeing systems after they are deployed. It frames this as a reporting problem that a standard checklist could help address across the lifecycle of healthcare AI tools.

For patients and clinicians using AI-assisted tools, the review suggests that it may be unclear who checks these systems once they are in use. Clearer reporting of human involvement could help users understand how and when a tool has been reviewed.

The takeaway

Clinicians are often involved in building clinical AI but seldom in keeping it under review. Asking developers to document who contributes, and how, is a practical step toward meaningful human oversight.

Further reading

For broader context on oversight of machine learning in medicine, see Artificial Intelligence.

Source note: This article includes information reported by Cyberdaily.

Live Poll

Would you feel comfortable with an AI tool helping decide your medical care?