AI Outperforms Humans in Predicting Breast Cancer

A study published in Radiology has found that artificial intelligence (AI) algorithms outperformed the standard clinical risk model for predicting a woman’s five-year risk of breast cancer. The clinical models currently used rely on various sources of information, which may not always be available. In contrast, AI algorithms can extract additional mammographic features, providing a more comprehensive assessment. The study analyzed data from negative screening mammograms conducted in 2016 at Kaiser Permanente Northern California.

Five AI algorithms were used to generate risk scores for breast cancer over the five-year period, and all of them performed better than the clinical risk model. Some algorithms were particularly successful in identifying patients at high risk of interval cancer, enabling early detection and appropriate follow-up. Combining AI with clinical risk models further improved the accuracy of cancer prediction. The study suggests that AI-based risk models could enhance personalized and precision medicine on a national level, as the risk scores could be integrated into radiology reports for patients and physicians.