Clairity Breast uses AI on a past mammogram to estimate five-year breast cancer risk, with caveats from oncologists

Key takeaways

Clairity Breast is the first FDA-authorized AI tool that analyzes a routine screening mammogram from the past 12 months to estimate a woman’s five-year breast cancer risk, and the 2026 NCCN guidelines now include AI image-based risk assessment as an option starting at age 35. Medical oncologist Tiffany Troso-Sandoval, MD, says the model flagged 41.5 percent of women who later developed cancer as high risk, compared with 22.9 percent for the Tyrer-Cuzick model, making it most useful for the 85 percent of breast cancer patients with no family history, especially those 35 to 50 with dense breasts. She cautions that the test is supplemental, covers only 2D images from specific Hologic systems, carries risks of false reassurance and unnecessary imaging and is best obtained through a patient’s own imaging center rather than a consumer upload.

More young women than ever are being diagnosed with breast cancer, with the fastest growth among women under the age of 50, according to statistics from the American Cancer Society.

Most women are aware of the recommendation to receive their first breast cancer screening with a mammogram at age 40, but even if they adhere to those guidelines, it may not be enough to provide full peace of mind. That’s because mammography often misses cancer in dense breast tissue, and about 50 percent of women over the age of 40 have dense breasts. However, a new FDA-authorized test using artificial intelligence may help women better understand their risk of developing breast cancer alongside consistent routine screening.

Clairity Breast by Everlywell, a digital health company known for its at-home lab testing kits, is the first FDA-authorized AI technology that analyzes a routine screening mammogram to estimate a woman’s risk of developing breast cancer over the next five years, using a qualifying mammogram from the past 12 months. According to Tiffany Troso-Sandoval, MD, a board-certified medical oncologist in Long Island, New York, the technology works by analyzing subtle patterns within mammogram images to provide additional personalized risk insights. Updated 2026 National Comprehensive Cancer Network guidelines include AI-based mammogram risk assessment as an option for identifying women at increased risk beginning at age 35.

According to Troso-Sandoval, this technology would be most useful for women without a family history of breast cancer. Those with a strong family history or a known mutation already have clear paths for assessing their risk, including Tyrer-Cuzick risk models, genetic testing and annual MRIs, and those who carry a genetic mutation would not be eligible for this test anyway.

“The woman this tool was built for is the one every questionnaire calls ‘average,’ because the questionnaire never looked at her tissue,” Troso-Sandoval says. “That is where most breast cancers come from.”

That’s not necessarily bad news, though, as about 85 percent of breast cancers occur in women with no family history of the disease, making them the biggest group who could stand to gain from new technology. However, medical experts want patients to separate the science from the delivery, as Clairity doesn’t detect or diagnose cancer or replace routine screening, Troso-Sandoval says. Rather, it’s intended to provide additional information women can discuss with their healthcare providers to guide screening and prevention decisions.

“The science is accepted enough that NCCN added AI image-based risk assessment to its 2026 screening guidelines, with a 1.7 percent five-year threshold defining increased risk starting at age 35,” she explains. “The delivery is where oncologists and radiologists are uneasy, as paying online, uploading an image and receiving a percentage from a clinician who has never met you is a different thing from risk assessment inside your own care.”

According to Troso-Sandoval, the biggest downsides of the test are named in the FDA authorization. They include a falsely low score that can make a woman decide to skip screening, and a falsely high one that sends her to get imaging she did not need. The FDA authorization also only covers 2D images from specific Hologic mammography systems and excludes synthetic 2D views, which means some women will discover they do not have a qualifying mammogram to use with it.

“The test is a deep learning model trained on hundreds of thousands of mammograms linked to five-year outcomes,” she says. “It reads pixel-level patterns in breast tissue that correlate with future cancer, patterns no human eye sees, and outputs a probability.”

According to Troso-Sandoval, this test is designed to be purely supplemental to regular screening. By its labeling, it’s a prescription device whose result is meant to be considered after the radiologist has read the mammogram and alongside the patient’s full history, all of which her provider should still review. In other words, even a good score does not mean a patient can get away with fewer mammograms, she adds, as no one’s odds of developing cancer will de-escalate based on an AI risk score.

“The difference is that a mammogram asks whether there is cancer in this breast today, while Clairity asks how likely a cancer is in the next five years,” she says. “A normal mammogram with a high score is not a contradiction, and a low score does not mean today’s mammogram was read correctly and that cancer is absent.”

What women do stand to gain, however, is reclassification, meaning someone the system considered to be of average risk may now qualify for supplemental MRI or ultrasound, earlier screening and a real conversation about risk reduction, including endocrine prevention for appropriate candidates.

“In the published research, the image-based model flagged 41.5 percent of women who went on to develop cancer within five years as high risk, compared with 22.9 percent for the Tyrer-Cuzick risk model,” Troso-Sandoval says. “That’s nearly double, and for the woman with no family history, that is the difference between cancer being found and being invisible.”

That said, she would still only recommend this add-on to women based on certain factors, such as being 35 to 50 with no family history, having dense breasts and being willing and able to act on the result.

“Ideally, she gets it through her own imaging center or physician rather than a consumer upload,” she adds. “The question I would ask her before she spends $249 on it is, ‘What will you do differently with the answer?’ If the answer is nothing, keep the money. If the answer is ‘I would finally get the MRI nobody offered me,’ it is worth it.”

Ultimately, it’s key for women to understand the score is a snapshot, not a lifetime label, and NCCN still calls for periodic reassessment, Troso-Sandoval says. Additionally, much like all other consumer health AI products and services, the test is sold as something that can provide empowerment, but the work of interpretation is handed back to the patient, who may not understand on her own that a good score doesn’t leave her in the clear.

“The empowered move is not buying the test. It is bringing the number to the person who holds the rest of your chart and asking what it changes,” Troso-Sandoval says. “The metaphor I use with patients is that the mammogram is the smoke detector, and Clairity is the weather forecast. You do not unplug the smoke detector because the forecast is clear.”

Emilia Benton is a Denver-based health and wellness journalist whose work has appeared in publications such as Runner’s World, SELF, and Women’s Health. She is also a 15-time marathoner, USATF Level 1-certified running coach, and an avid traveler.