Artificial intelligence in Sweden’s national breast cancer screening program proved more effective than traditional methods, detecting more clinically significant cancers without increasing false positives. Full results from a 2023 trial, now published in The Lancet, confirm these benefits from the Mammography Screening with Artificial Intelligence (MASAI) study.
Researchers from Lund University and institutions in Sweden, Norway, Denmark, and the Netherlands led the effort. Women screened with AI support showed lower rates of aggressive or advanced breast cancers over the next two years. Interim findings in August 2023 already indicated a 20 percent rise in cancer detection and a 44 percent drop in radiologists’ workload.
The complete data revealed a 12 percent reduction in interval cancers, those found between screenings after a negative result. Lead author Dr. Kristina Lang, a breast radiologist and clinical researcher at Lund University, called it the first randomized controlled trial on AI in breast cancer screening and the largest overall for AI in cancer detection. She noted that AI boosts early identification of relevant cancers, helping curb aggressive cases that are spotted between appointments.
From April 2021 to December 2022, the trial randomly assigned over 105,900 women to either AI-supported screening or standard double reading by radiologists. The AI system drew on training data from more than 200,000 exams across over ten countries. In a two-year follow-up, the AI group had 1.55 interval cancers per 1,000 women versus 1.76 in the standard group.
Detection at screening reached 81 percent in the AI arm, up from 74 percent in standard screening, a nine percent improvement. False positive rates stayed comparable at 1.5 percent for AI and 1.4 percent for standard. Lang suggested a broad AI rollout could ease radiologists’ burdens while catching more early-stage and aggressive cancers.
