By January Lopez, M.D.
For the last few years, artificial intelligence (AI) has been proving itself as an effective tool in breast cancer screening. An ever growing list of research studies have shown that using AI in mammography can safely and efficiently reduce patient wait times while reducing the pressure on in-demand radiologists.
This is significant, as early detection greatly improves patient outcomes and long term survival. Imagine how much more significant it would be if AI could predict future cancer.
A study published this summer in the journal Radiology, offered a sneak peek into what I expect will be a growing body of evidence about the use of AI in determining five-year breast cancer risk from the mammograms of women who do not yet have the disease. According to the study, the AI tools seem to be able to better predict a woman’s risk than the tools we have now.
All women are at risk of developing cancer. We know that some women are at elevated risk due to lifestyle or genetics. As it turns out, there may be hidden information inside a mammogram that predicts a woman’s future potential for developing breast cancer.
Current risk assessment is calculated using the Breast Cancer Surveillance Consortium (BCSC), a model that predicts risk using age, race or ethnicity, first-degree family history of breast cancer, number of prior benign breast biopsies, and mammographic breast density.
The study showed that the women with the highest BCSC risk scores at the start of the study accounted for 21.1% of all cancers over the next 5 years. In comparison, the women with the highest AI risk scores accounted for 24% to 28% of all cancers. More telling – combining the BCSC and the AI scores led to even greater accuracy.
Early detection programs can help women take control of their health by encouraging lifestyle changes, frequent screenings and surveillance to help stave off breast cancer or catch the disease early.
But the study did not include the kind of women who are usually invited to participate in such programs. The study did not include women with previous breast cancers or genetic mutations. A quarter of them were under 50 at the start of the study and 87% had no first degree family history of the disease. The women in this study were truly the ones for whom breast cancer was “something that happened to other people.”
That is what is so thrilling about the potential to use AI in radiology as a predictive tool of health. While survival rates are increasing, we still lose 42,000 women and 500 men to breast cancer every year. Unlocking the power of prediction could save lives.
Most AI algorithms trained to read mammograms are being used to detect existing cancers. Let’s put them to work in the service of prevention.
Dr. January Lopez is a radiologist and medical director of breast imaging with Hoag Memorial Hospital Presbyterian.
