Study Explores Simple Way to Predict Breast Cancer Using Images and Genes - Video

What if your genes and your mammogram could team up to predict your breast cancer risk more accurately?

A new study, published in the British Journal of Cancer, suggests that combining artificial intelligence with genetic information could make breast cancer prediction smarter and more personalized. Researchers looked at a deep learning model called Mirai, which analyzes mammogram images to estimate a woman’s risk of developing breast cancer within the next five years.

Traditionally, doctors rely on tools like the Gail Model, which uses factors such as age, family history, and reproductive history. But these methods don’t always capture the full picture.

In this study, scientists added something new: a polygenic risk score (PRS). In simple terms, this is a number based on many small genetic changes in your DNA that together influence your cancer risk. Think of it like adding up tiny risk signals from your genes.

The study included over 900 women. Researchers compared how well the AI model worked on its own versus when it was combined with genetic data. The results showed a clear improvement. The AI model alone had moderate accuracy, but when combined with PRS, it became better at identifying who was more likely to develop breast cancer.

Even the traditional Gail model improved significantly when genetic information was added—but the AI + genetics combination still performed the best overall.

What does this mean in real life? Instead of using a one-size-fits-all approach, doctors could soon offer more personalized screening plans. Women at higher risk might be monitored more closely or start screening earlier, while those at lower risk might avoid unnecessary tests.

It’s important to note that this approach is still being studied and is not yet standard practice everywhere. But it shows a promising future where your scan and your genes work together—helping detect risk earlier and potentially saving lives through smarter prevention.

REFERENCE: Azam, S., Lamb, L.R., Eliassen, A.H. et al. Performance of an image-only deep learning breast cancer risk model with the addition of a polygenic risk score. Br J Cancer (2026). https://doi.org/10.1038/s41416-026-03415-z