A new deep learning model, developed by a team from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Massachusetts General Hospital (MGH), is attempting to detect patients at risk of breast cancer at an early stage, before the development of the disease, avoiding later diagnosis.

The approach is aimed at highlighting those subtle patterns of breast tissue, future cancer marker signals, that are difficult to intercept by a doctor’s observation. This tool, learned from more than 90,000 mammograms, is able to place 31% of all cancer patients in the highest-risk category, compared to the 18% obtained by existing models.

Furthermore, the model by MIT and MGH model could guarantee an equivalent application for all women, contrary to what is already available, developed on populations of white women. It is worth noting that African American women are 43% more likely to die due to breast cancer than white women.

The scientific community is now grappling with the screening dilemma “When to start and how often”; based on this model, the perspective could change course. “Rather than taking a one-size-fits-all approach, we can personalize screening around a woman’s risk of developing cancer”, says MIT Professor Barzilay. “For example, a doctor might recommend supplementary MRI screening for women with high model-assessed risk”.

Precision medicine requires precise diagnosis: the is the road that has been taken, and in the near future, systems such as this one based on an AI-Model could make prevention and screening ultra-customized for a variety of diseases, including cancer.