The 510(k) submission covers a mammography-based model designed to estimate five-year breast cancer risk from screening images.
Lunit, a provider of AI for cancer diagnostics and precision oncology, submitted a 510(k) premarket notification to the US Food and Drug Administration (FDA) for Lunit INSIGHT Risk, a mammography-based five-year breast cancer risk prediction model. Potential clearance is expected in 2026.
Lunit INSIGHT Risk estimates a woman’s five-year breast cancer risk directly from screening mammograms (either synthetic or digital mammography) without the need for patient questionnaires. The technology was originally developed at Washington University School of Medicine in St. Louis by Graham Colditz, MD, DrPH, and Dr Shu (Joy) Jiang, and later acquired by Lunit.
The company claims it is the first AI solution to generate SEER-calibrated five years absolute risk estimates for US women. For example, a 5% risk indicates that five of 100 women with similar risk profiles are expected to develop breast cancer within the next five years.
The model’s performance has been validated in two peer-reviewed studies published in JAMA Network Open and JCO Clinical Cancer Informatics, showing discriminative performance with five-year AUCs up to 0.80 across diverse screening populations in the US and Canada. These studies also showed consistent performance regardless of age, race, or breast density, supporting its potential for equitable clinical adoption.
“Image-based risk assessment represents an important step toward making breast cancer prevention more precise and more accessible,” says Colditz, professor of Washington University School of Medicine, in a release. “Our research has shown that mammography-derived five-year absolute risk can be both highly discriminative and well calibrated to US disease incidence rates, allowing clinicians to identify women who may benefit most from supplemental screening or preventive strategies. With Lunit advancing this technology toward regulatory review, we have an opportunity to bring evidence-based, personalized risk prediction closer to routine clinical practice.”
Advancing the Tech
The model received FDA Breakthrough Device Designation last April, and Lunit had been participating in the FDA Total Product Lifecycle Advisory Program, which enables more structured and frequent engagement with the agency during development and review. The program also supports discussions with non-FDA stakeholders to expedite market adoption, clinical use, and patient access.
“Submitting our first 510(k) for an image-based risk prediction model is an important milestone for Lunit and for the advancement of personalized breast cancer screening,” said Brandon Suh, CEO of Lunit, in a release. “By delivering absolute, guideline-aligned five-year risk estimates directly from routine screening images, Lunit INSIGHT Risk aims to support earlier and more informed decisions in preventive care. We believe this approach can help health systems implement practical risk-stratified strategies as screening recommendations continue to evolve.
Lunit INSIGHT Risk is designed to integrate with the company’s broader breast health portfolio, including Lunit INSIGHT MMG and DBT detection models and Volpara Risk Pathways, creating a comprehensive foundation for risk-informed care workflows that span assessment, imaging, reporting, and longitudinal follow-up.
ID 385738869 | Imaging © SlavkoSereda | Dreamstime.com
