The topline. Most women with early breast cancer already get an MRI before treatment. A newly cleared AI tool asks what else that scan knows. The FDA has granted 510(k) clearance to SimBioSys TumorSight Risk, software that analyzes a pre-treatment breast MRI with pathology and clinical data to estimate recurrence risk at five and ten years. It requires no new procedure or tissue shipment, and it is designed to sit alongside the oncologist’s existing tools rather than replace any of them.

Key facts

The device is cleared for adult women with hormone receptor positive, HER2 negative stage I to IIIA breast cancer, about 75 percent of US breast cancers. The workflow runs in sequence with the company’s previously cleared TumorSight Viz: a pre-treatment contrast enhanced MRI becomes a tissue segmentation map, which passes into TumorSight Risk. The software combines imaging values with age, race, tumor stage, nodal stage, and tumor grade, producing a score from zero to 100: low at 20 or below, intermediate above 20 to 35, high above 35. It is cleared for GE and Siemens 1.5T and 3T scanners.

Clearance rests on a blinded, prospectively designed retrospective validation of 2,129 patients from Mayo Clinic, Emory University, Stanford Medicine, and Mercy Medical Center in Cedar Rapids. Patients classified high risk had a 3.3 fold greater risk of recurrence than those classified low risk. Estimated five year recurrence risk ranged from 2.2 percent for the lowest scores to 13.6 percent for the highest. The score remained significantly associated with recurrence free survival after adjustment for standard prognostic information. The five year negative predictive value was 97.0 percent and the ten year value 91.4 percent, while positive predictive values were 9.8 percent at five years and 26.4 percent at ten years, a reminder that a high score raises concern without confirming recurrence.

Clinical relevance

This is AI in its most useful clinical posture: augmentation, not adjudication. The information comes from imaging the patient already had, it arrives earlier than a tissue based genomic assay, and the clearance positions it as one input among others, to be considered alongside clinicopathologic factors. For an oncologist weighing treatment intensity or monitoring frequency, an imaging native risk estimate adds a second lens, and its high negative predictive value is most valuable where it reassures: a low score can support a less intensive path.

A 510(k) clearance means the device was found substantially equivalent to a legally marketed predicate, here ArteraAI Breast, not that it outperformed existing tests in a head to head trial. The validation was retrospective, the company notes Mayo Clinic has a financial interest in SimBioSys, and the score’s positive predictive value is modest, so a high score is a prompt for closer attention, not a verdict. Clinicians should read it the way the label intends: one more piece of evidence in the room, with the decision still theirs.

97.0%

That is the five year negative predictive value: among patients the tool did not flag, 97 out of 100 were recurrence free at five years. For a prognostic tool, that is the number that earns its place in the clinic. Ruling out is often more valuable than ruling in, and a tool that can say low risk with that confidence, from a scan already taken, gives the team room to spare a patient the heavier path.

What we’re reading

  • U.S. Food and Drug Administration: the 510(k) clearance letter and summary for TumorSight Risk.
  • SimBioSys: the October 2026 clearance announcement and validation summary.
  • American Society of Clinical Oncology Annual Meeting 2026: the validation presentation.