Observed arrival · 2026-11-20
Algovigilanza’s proposed safety framework for clinical AI
An Italian-language proposal for monitoring and classifying harms from clinical algorithms after they are put into use.
- For
- Clinical AI safety and incident-reporting teams
- Worth noticing
- The page calls the framework a proposal and says inter-rater reliability, usability, and effects on safety have not been measured.
Field notes
The proposed record treats a single incident as potentially affecting patients, professionals, and the organization, and keeps outcomes, mechanisms, and indicators distinct. Its evidence section cautions that published examples are heterogeneous and usually do not measure harm directly. One cited sepsis-model evaluation covers 38,455 admissions and reports 33% sensitivity at the threshold assessed; the page presents those figures as signals of risk and alert burden, not proof of patient harm.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue