AI Tool Predicts Breast Cancer Risk Years Ahead Using Routine Mammograms

Breast cancer screening saves lives but still relies largely on uniform schedules despite wide differences in individual risk. This one-size-fits-all approach can miss cancers in higher-risk women while leading to unnecessary tests in those at low risk. Researchers now report an artificial intelligence (AI) model that uses routine screening images to estimate a woman’s four-year breast cancer risk and support more personalized surveillance.

The BRAIx risk score is an AI model developed by an international team led by St Vincent’s Institute of Medical Research (Melbourne, Australia) as part of the BRAIx program. The approach analyzes information already contained in digitized mammograms to estimate a woman’s likelihood of developing breast cancer within four years.

The system focuses on imaging-derived patterns rather than relying solely on age, breast density, or family history. By mining features present in standard screening images, it generates an individualized near‑term risk score. The method is designed to work with data available from existing screening workflows without additional testing.

Model development used mammograms from nearly 400,000 women screened at BreastScreen Victoria. Performance was then tested on almost 96,000 Australian women. Findings were confirmed in an independent Swedish population of more than 4,500 women.

Results showed that the tool identified a high‑risk subgroup with notable precision. Nearly one in ten women ranked in the top 2% by the BRAIx risk score were diagnosed with breast cancer within four years despite a prior all‑clear. The study was published on March 4, 2026, in The Lancet Digital Health.

Investigators report that the BRAIx risk score could help customize screening intensity by escalating surveillance for those at higher near‑term risk and reducing frequency for those at very low risk. Additional studies are planned before routine clinical adoption. The program is funded by the Medical Research Future Fund.

“Our AI risk work offers the potential for the most significant step change in breast cancer mortality since screening was introduced in Australia in the early 90s,” said Associate Professor Helen Frazer, St Vincent’s BreastScreen Melbourne. “Given the potential we’re seeing with AI in the BRAIx program – its accuracy, and the information it provides clinicians – zero deaths from breast cancer is a real and genuine long-term goal.”

Related Links
St Vincent’s Institute of Medical Research