The tool analyzes morphological parameters from standard pathology slides that doctors already use to diagnose cancer. This approach could help doctors determine whether and which chemotherapy is likely to work for each patient, and importantly, does not require any additional procedure or cost.
By identifying which patients are more likely to respond to a specific chemotherapy early on, the tool could help maximize therapeutic benefits, spare patients from unnecessary side effects and allow them to explore clinical trials sooner.
“From the time of diagnosis, this AI tool can help identify which treatments are most likely to benefit ovarian cancer patients,” said Martina Bazzaro, PhD, an associate professor at the University of Minnesota Medical School and a Masonic Cancer Center researcher. “It can reduce uncertainty early in the treatment planning process, avoid therapies that are unlikely to help, and lessen both the emotional and financial burden on patients and the health care system. We are now moving toward a clinical trial to integrate this tool into clinical care, harnessing cutting-edge AI to guide treatment decisions and improve outcomes, while supporting more efficient and cost-effective healthcare delivery.
"The use of AI in medicine, and in oncology in particular, is evolving rapidly,” said Emil Lou, MD, PhD, FACP, a professor at the University of Minnesota Medical School and gastrointestinal oncologist with the Masonic Cancer Center and M Health Fairview. “Our research developed an AI tool for identifying Tumor-Stroma Proportion - a promising measure from tumors at time of first diagnosis - with a high level of agreement compared to review by expert pathologists. This step is a promising example of how we can move things forward into the clinic more quickly than ever before."
Future studies will aim to integrate the AI tool into clinical trials within the next six months to further evaluate its real-world impact and integrate the technology into health systems so results can be delivered automatically to care teams.
The research was funded by Minnesota Ovarian Cancer Alliance, Randy Shaver Community Fund and the American Cancer Society.
Aggarwal A, Madill M, Jana M, Pathak T, Starr TK, Winterhoff B, Tessier KM, Erickson BK, Nelson AC, Lou E, Madabhushi A, Bazzaro M.
STAR (stroma-tumor AI risk) assessment: association of AI-derived tumor-stroma proportion with patient survival provides added prognostic value beyond KELIM in epithelial ovarian cancer.
BJC Rep. 2026 Feb 6;4(1):4. doi: 10.1038/s44276-026-00205-1