New research shows yet another way Artificial intelligence is entering the medical field — and may improve existing practices for predicting breast cancer risk.
The study, published Tuesday in the peer-reviewed journal Radiology, found that AI algorithms outperform standard clinical risk models that predict risk over a five-year period. breast cancer.
Calculations of breast cancer risk in women typically use risk models, such as the Breast Cancer Surveillance Consortium (BCSC) Clinical Risk Score, which uses self-reported information and other patient information, including age, family history, etc.
“Clinical risk models rely on collecting information from a variety of sources, and information is not always available or collected,” says Kaiser Permanente, a research scientist and practicing radiologist at Northern California. Principal Investigator Dr. Vignesh A. Arras said in a news release. “Recent advances in AI deep learning have made it possible to extract hundreds to thousands of additional mammography features.”
In this retrospective study, thousands of mammograms were analyzed and five AI algorithms generated a five-year breast cancer risk score. These scores were then compared to each other and the BCSC clinical risk scores were compared.
“All five AI algorithms performed better than the BCSC risk model when predicting breast cancer risk from years 0 to 5,” said Arasu. “This strong predictive performance over five years suggests that AI can help identify both missed cancer and breast tissue features and predict future cancer development.”
While some centers are already using AI to detect cancer through mammography, the findings suggest that AI could be an important tool in helping patients’ future risk scores. Yes – according to the release, it takes a few seconds to generate the AI.
“AI for cancer risk prediction offers us the opportunity to personalize all women’s care that is not systematically available,” said Arras. “This is a tool that will help us deliver personalized precision medicine at the national level.”
