Randomized Study of AI-Assisted vs Standard Screening Mammography for Breast Cancer Detection and Recall in Adults

What this Trial Is About

Researchers are conducting a randomized controlled trial to compare the outcomes of screening mammography exams interpreted with and without the help of an FDA-cleared artificial intelligence AI decision-support tool in real-world U.S. settings. The study aims to determine if using AI improves breast cancer detection and recall rates during screening. This trial includes all adult patients undergoing screening mammography and all interpreting radiologists across six regional health systems. During the trial, each 3D screening mammogram will be randomly assigned to either the intervention group, where radiologists receive assistance from the AI decision-support tool, or the usual care group, where radiologists interpret the images alone. Randomization happens at the exam level immediately after image acquisition, and patients returning for screening in the second year will be re-randomized. Radiologists will see AI information only when it is available during interpretation, but they maintain full control over their final diagnosis. Participants will undergo routine screening mammograms, with data collected on cancer detection rates within 90 days and recall rates over about one year. Additional measures include false-positive rates, interval cancer rates, and trust in AI over several years. The study plans to assess these outcomes across approximately 400,000 screening exams, with monitoring done through the participating health systems. Radiologists and patients will contribute perspectives on AI use in medical imaging during the trial period.

A Trial Comparing Screening Mammography With and Without Assistance From Artificial Intelligence for Breast Cancer Detection and Recall Rates in Adult Patients