Researchers at UC San Francisco and UC Berkeley have developed an AI-driven workflow to significantly reduce wait times for women with abnormal mammograms. The AI model, called Mirai (developed by UC Berkeley data scientist Adam Yala), analyzes screening mammograms to identify subtle patterns and predict cancer risk more accurately than a physician working alone. After being trained on hundreds of thousands of mammograms linked to patient outcomes, the model was applied to more than 4,100 screening mammograms at Zuckerberg San Francisco General Hospital and Trauma Center. Mirai determined that about 12.7% of screened patients were high-risk. For these women, the AI-guided workflow allowed them to get an interpretation of their mammogram immediately after the scan, followed by additional diagnostic imaging — and in some cases, a biopsy — all on the same day. Without AI, the wait for a diagnostic evaluation could take several weeks. For those ultimately diagnosed with breast cancer, the average wait for a biopsy was reduced from more than two months to fewer than 10 days. Importantly, Mirai does not replace radiologists or make diagnoses on its own. Instead, it serves as a triage tool, helping physicians quickly identify patients who can benefit most from accelerated care. This is a powerful example of how AI can collaborate with doctors to deliver more personalized and timely care.
About the Author
