In this randomized, controlled trial, women aged 40–80 eligible for mammography screening were assigned to either AI-supported screening or standard double reading without AI. The AI system provided a malignancy risk score to triage examinations, and the study aimed to evaluate the impact of AI on cancer detection rates, recall rates, false positives, and screen-reading workload. The clinical safety analysis will determine whether AI-supported screening meets the minimum acceptable cancer detection rate of more than 3 per 1000 participants and will continue to monitor interval cancer rates.
Creator
Lund University, Skåne University Hospital, Cancer Registry of Norway, The Arctic University of Norway