Colon cancer, a leading cause of cancer death globally, presents symptoms like abdominal pain and changes in bowel habits. Researchers at Harvard Medical School and National Cheng Kung University have developed an AI model called MOMA (Multi-omics Multi-cohort Assessment) to improve treatment decisions for colorectal cancer patients. This tool analyzes tumor sample images to predict tumor aggressiveness, survival probabilities, and optimal therapies, often outperforming human pathologists. MOMA could significantly assist clinicians, especially in resource-limited areas lacking advanced pathology services.
Trained on data from nearly 2,000 patients across diverse cohorts, MOMA examines patient demographics, cancer stages, and tumor genomic profiles. It accurately forecasts patient outcomes and therapies by detecting visual patterns that elude human evaluation, such as cell density and tumor microenvironment characteristics.
The researchers emphasize MOMA is intended to augment human expertise rather than replace it, suggesting the need for a prospective, randomized trial before clinical implementation. The model’s transparent reasoning enhances clinician confidence in its predictions. As science evolves, the model will be updated to ensure accuracy, making it a valuable asset in cancer management and a potential lifesaver in the fight against colorectal cancer.
