Look beyond
the headline.
A system that performs well on a test can still fail in a clinic. Patient populations, workflows, professional oversight and the consequences of a missed diagnosis all affect responsible use.
- 01Demand evidence from the intended clinical setting.
- 02Inspect performance across relevant patient groups.
- 03Preserve human review and a route to appeal.
What the figure
can tell us.
WHO’s core principles for the ethical use of AI in health, including human autonomy, safety and accountability.
Read WHO, 2021 ↗Read the source for its methods and limits. A published estimate, observed result and future projection answer different questions. The research figures describe this topic; they are not learner scores.
A benefit.
A consequence.
What it can give
Decision support and administrative assistance may give clinicians more time and extend access to expertise when validated for the setting.
What it can cost
Uneven training data, automation bias and uncertain accountability can expose patients to harm. A tool can be accurate overall while performing poorly for a particular group.
The judgement depends on context. Which effects matter most, who experiences them, and what evidence would change the decision?