Changes in AI mammogram risk scores over time help predict future breast cancer

The study published in Radiology used artificial intelligence (AI) and deep learning models to develop dynamic breast cancer risk scores from screening mammograms.
Unlike traditional models, these AI-generated scores are based directly on image data rather than on limited factors such as breast density.
The study revealed that these AI risk scores evolve over time and exhibit notable differences in trajectories between women who develop breast cancer and those who do not.
Researchers found that among women who developed cancer, these risk scores started to increase significantly about six years before diagnosis, progressively rising up to a steep incline just before the disease was detected.
This progressive increase was notably absent among the cancer-free control group, where the AI risk scores remained stable during the study period.
The findings highlight the potential of AI risk scores to identify women predisposed to breast cancer who might not otherwise have been flagged as high risk because they do not have a significant family history or genetic mutations.
The study supports the view that these image-based AI risk models serve as dynamic biomarkers that could pave the way for more personalized, risk-based preventive strategies for breast cancer.
Results from the study may help to reduce existing disparities in screening performance across different patient demographics by offering a risk assessment tool based purely on imaging data.

You may have missed