
Artificial intelligence (AI) algorithms have shown superior performance in predicting the five-year risk of breast cancer compared to standard clinical risk models, according to a large study analyzing thousands of mammograms. The study, conducted at Kaiser Permanente Northern California, utilized data from screening mammograms performed in 2016. Five AI algorithms, including academic and commercially available models, generated risk scores that outperformed the Breast Cancer Surveillance Consortium (BCSC) risk model. AI algorithms identified missed cancers and breast tissue features associated with future cancer development. Combining AI and clinical risk models further enhanced cancer prediction. The findings suggest that AI-based risk models using mammograms as the data source offer practical advantages in assessing breast cancer risk.
Like
Save
Share