The National Comprehensive Cancer Network has added AI-based mammogram risk assessment to its 2026 breast cancer screening guidelines. Some imaging centers will now run a computer model over each patient's routine mammogram and assign a five-year cancer risk score.
Women whose score reaches at least 1.7% are considered at increased risk. The guidelines link that threshold to supplemental imaging, risk-reduction counseling and periodic reassessment, and expand risk identification starting at age 35.
The visit itself does not change. The same image is taken, and it does not take longer. What is new is that pixel-level patterns in the mammogram, patterns the human eye cannot see, are added to family history and other factors to estimate future risk.
What the numbers behind the guideline show
In a multicenter analysis of 245,232 screening mammograms presented at the RSNA 2025 annual meeting, women the AI flagged as high risk had a five-year breast cancer incidence of 5.9%. Women it categorized as average risk had a 1.3% incidence over the same period.
Also read: These vegetables make it easier to eat more fiber
That is roughly a 4.5-fold difference. Breast density, the risk factor most centers report today, produced a much narrower split in the same dataset: 3.2% incidence in dense breasts versus 2.7% in non-dense breasts.
"The model is able to detect changes in the breast tissue that the human eye can't see," said Constance D. Lehman, MD, PhD, professor of radiology at Harvard Medical School, in the RSNA release. Lehman is also the founder of Clairity, the company behind the algorithm.
How the algorithm actually reads a mammogram
The tool named in the NCCN update, Clairity Breast, received FDA De Novo authorization on June 2, 2025. It is the first AI model cleared for predicting five-year breast cancer risk directly from a standard bilateral 2D mammogram, and it was validated across more than 77,000 mammograms from five geographically distinct screening centers.
It works from the same image radiologists already review, generating a personalized score without requiring genetic testing or a detailed family-history questionnaire. That matters because most women who develop breast cancer have no strong family history or inherited mutation, and traditional questionnaire-based risk models can miss them.
Also read: FDA approves the first weight-loss pill you can take any time, without food restrictions
What it means at your next appointment
If your imaging center has adopted the AI tool, you should receive a numerical five-year risk estimate with your mammogram result. A score below 1.7% falls in the average-risk band, and routine screening continues.
A score at or above 1.7% moves you into the increased-risk group under NCCN's new definition. That is not a diagnosis and does not mean cancer is present or likely soon. It means your provider should discuss extra steps, which can include supplemental MRI or ultrasound, risk-reduction medication and periodic reassessment.
If you do not know whether your center offers AI risk scoring, ask when you book your next mammogram. Availability is uneven, and Clairity Breast is only beginning to appear in select U.S. hospitals. Lehman told TODAY that the test can be offered for less than $200, though insurance coverage varies.
A high score is not a certainty, and a low one is not a guarantee. Anyone noticing a new lump, nipple change, skin dimpling or unexplained breast pain should contact their doctor regardless of any risk number on the screen.
Also read: Your immune system remembers obesity for years, even after the weight is gone
Also read: Study links neighborhood in youth to heart health later in life
This article is made and published by Jesper Bengtson, who may have used AI in the preparation.