A new AI model developed by Technion-Israel Institute of Technology researchers, alongside international medical centers, predicts both breast cancer recurrence risk and the likelihood that a patient will benefit from chemotherapy.

The model analyzes routine pathology slides at diagnosis, offering a faster, more accessible alternative to genomic tests like Oncotype DX.

Validation involved thousands of patients, including data from Rambam Healthcare Campus and large randomized breast cancer studies, showing consistent performance across diverse populations and equipment.

Researchers aim for prospective validation before commercial deployment and envision forming a startup to scale accessible, faster tests globally.

Future directions include extending the model to other treatments and cancer types, with year-long clinical trials planned in countries such as India, the Philippines, and Brazil.

Key figures include Dr. Gil Shamai, who leads the Geometric Image Processing Laboratory, collaborating with Prof. Ron Kimmel, Prof. Dvir Aran, and hospitals across Israel, the US, and Australia.

The model derives a numerical score from visual tissue patterns that correlate with recurrence risk and treatment sensitivity, without requiring additional tissue or lab processing.

The study, published in The Lancet Oncology, marks the first AI model validated in a large randomized clinical trial for predicting chemotherapy benefit from pathology images.

Compared to genomic tests, the AI approach is fast (measured in minutes), eliminates waiting periods, and can be deployed in any digital pathology lab with internet access, potentially lowering costs and expanding access in lower-income regions.