Clairity’s FDA-authorized AI can now tell a woman her breast cancer risk years before a diagnosis
For most of her career, Dr. Connie Lehman stood in front of a light box, reading mammograms one film at a time. What she kept noticing was that no two looked alike. That observation, small as it seemed, would eventually lead her out of the hospital and into a startup. Today, as founder and CEO of Clairity and a professor of radiology at Harvard Medical School, she is making a case that the mammogram has always held more information than medicine ever bothered to extract.
What a mammogram has always known
Dr. Lehman traces the idea back to 1967, when a physician named John Wolf first published a paper connecting mammogram patterns to future cancer development. “Those observations, we couldn’t really do a lot with them, until we had the power of AI,” she said. That power arrived in the form of deep learning, a branch of artificial intelligence that taught computers to read images the way a radiologist might, only faster and at a scale no human team could match.
Clairity trained its model on hundreds of thousands of mammograms, split between women who developed breast cancer within five years and women who did not. The model learned to distinguish between the two. No additional patient data was needed. “It turned out that adding in other information about the woman didn’t improve the predictive power of our image-based model,” Dr. Lehman said.
Why traditional risk models failed Black women
The technology arrives at a moment when the limits of older risk assessment tools are hard to ignore. “Our traditional methods of assessing risk were built largely on European Caucasian women, and we’ve known for decades they do not translate, they do not work for Black women,” Dr. Lehman said. “It’s 2026, we’ve known this for 50 years, and no real change has been made.”
Clairity built its data consortium differently. The company pulled mammogram data from across the country, from Texas to Georgia, and included Hispanic, Asian and Black women in both its training set and its FDA clinical validation studies. “We will not repeat the sins of the past,” she said.
The shift from reaction to prevention
The National Comprehensive Cancer Network updated its 2026 guidelines to include AI-based mammogram risk assessment, a move Dr. Lehman called a turning point. Screening recommendations will now be tailored to the individual woman rather than applied as a blanket age threshold. “We’re going to update our guidelines and include new information so women can benefit now, not in the next generation,” she said.
For women, the practical difference is significant. A mammogram no longer just answers whether cancer is present today. It can now signal whether cancer is likely in the next five years, giving doctors and patients time to act. “We’re going to be proactive, not reactive,” Dr. Lehman said. “We’re going to find that risk early, and we’re going to reduce that risk.”
What Clairity wants women to do now
The technology is currently available in Boston, with launches planned in Georgia and Colorado. Women can join a waiting list and follow updates at clairity.com. “I wanted AI to be in Clairity,” Dr. Lehman said.
Her message to women is straightforward. “I want every woman to know her risk. We have new ways to assess risk that we never had before, and that information is empowering.” For the women who have long been told their family history puts them in the clear, that reassurance may no longer be enough. “Most women diagnosed with breast cancer have no family history,” Dr. Lehman said. “We’re missing easily 75% to 80% of women that will be diagnosed.”
The goal, she said, is to stop waiting for cancer to appear before taking action. “That’s the new world.”