This AI Model Predicts Breast Cancer Years Before Humans

Added: 2026-05-08

[00:00:02]Breast cancer is the most common cancer among women in the Western world, affecting nearly 1 in 9 women over the course of her lifetime.

[00:00:11]Yet, despite decades of progress, breast cancer screening still asks a limited question what do we see today?

[00:00:22]We know that in order for a radiologist to actually spot the cancer, this cancer needs to be large enough that human eye can see it.

[00:00:31]We also know that cancer takes a very long time to develop.

[00:00:38]Mirai is an AI model that takes as an input a mammogram, and it can predict the risk of a patients to develop breast cancer in the next five years.

[00:00:49]Trained on tens of thousands of mammograms, including those from women who later developed breast cancer, Mirai learns Mirai learns which tissue characteristics are associated with increased future risk.

[00:01:01]Given these pairs of images and the outcomes for the next five years after the mammogram was taken, We can teach the machine to correlate these changes in the image with the likelihood of developing breast cancer.

[00:01:20]We’re used to looking at mammograms as snapshots of the present Mirai looks at the same images and looks into the future.

[00:01:27]It's a totally different way of looking at mammograms.

[00:01:30]It's actually an entire paradigm shift in the way we screened for breast cancer.

[00:01:36]Most women are screened based on age rather than individual risk.

[00:01:41]This means some women are monitored too late, while others undergo years of screening they may not need.

[00:01:48]In the era of personalized medicine.

[00:01:51]We should also be able to offer personalized screening based on a woman's level of risk, rather than the current situation of age-based, one size fits all screening for the entire population.

[00:02:03]Mirai was validated on 2 million mammograms in multiple countries, demonstrating reliable performance in real world clinical settings.

[00:02:14]Local validation is essential and ensures that the technology is appropriate for the population that we serve and helps build trust among clinicians and patients.

[00:02:24]Like in all areas of our life Knowledge is power.

[00:02:28]If we know that we are at risk there are lots and lots of things that we can do today to mitigate the risk or to treat the disease as soon as possible.

[00:02:39]We know the treatments are becoming better every day, but we also know that the sooner the patient is entering the treatment round, the better are the outcomes.

[00:02:49]Mirai can help us reshape breast cancer screening, improving early diagnosis and ultimately the outcomes of the disease.