Hours Instead of Weeks: NYU's AI Reads Pathology Slides to Predict Breast Cancer Recurrence at Fraction of Genomic Testing Cost

Breast cancer patients facing treatment decisions may soon receive critical recurrence risk information in hours rather than weeks, thanks to an artificial intelligence system developed by researchers at New York University that analyzes standard pathology slides combined with routine clinical data.

The AI-based test, published in Nature Communications, was evaluated across 15 patient populations in seven countries involving more than 3,500 patients. Led by Krzysztof J. Geras—a visiting scholar at NYU Center for Data Science and co-founder of Ataraxis AI—the research demonstrates that multi-modal AI can match or outperform existing genomic tests while delivering results faster and at lower cost.

Bridging the Gap for Hard-to-Treat Cancers

Current genomic testing primarily serves hormone-receptor-positive breast cancer patients, leaving significant populations without reliable recurrence prediction tools. The NYU system breaks new ground by performing well for triple-negative and HER2-positive breast cancers—subtypes that currently lack genomic testing options and are often more aggressive and difficult to treat.

The technology combines pathology slide analysis with routine clinical information including tumor stage, patient age, and hormone-receptor status. Using self-supervised pretraining—a technique that allows AI to learn data representations before tackling specific prediction tasks—the system was evaluated using C-Index for predictive discrimination and Hazard Ratio for comparative risk assessment.

Co-author Yann LeCun, a professor of computer science and data science at New York University and a pioneer in deep learning, contributed to the development of the multi-modal approach that leverages existing pathology infrastructure rather than requiring new tissue processing methods.

Clinical and Economic Implications

For healthcare providers, the accelerated timeline—from weeks to hours—enables faster chemotherapy decisions while preserving precious tissue samples for additional testing as new biomarkers emerge. In resource-limited settings where genomic testing remains cost-prohibitive, the technology could expand access to precision oncology decision-making.

The system's ability to work with standard pathology slides means hospitals wouldn't need to invest in new tissue processing infrastructure, potentially accelerating adoption. The tissue preservation advantage offers patients additional benefits beyond faster results—saved samples remain available for future testing as cancer research advances.

The research emerges alongside Ataraxis AI, a commercial entity in which multiple paper authors hold equity and through which NYU maintains financial and intellectual property interests.

Path to Clinical Deployment

Despite promising results across international patient populations, the researchers acknowledge that clinical deployment requires evaluation in completed randomized clinical trials. This additional validation step will be crucial for regulatory approval and integration into clinical guidelines.

As artificial intelligence continues transforming medical diagnostics, this multi-modal approach demonstrates how combining traditional pathology with machine learning can extend precision medicine benefits to previously underserved patient populations while reducing costs and turnaround times.

Source: Futurity.org