AI Can Detect Colorectal Cancer Using Gut Bacteria

(University of Geneva, August 27, 2025)

Researchers at the University of Geneva and Geneva University Hospitals, led by Dr. Mirko Trajkovski and Dr. Matija Trickovic, have developed a revolutionary non-invasive method for colorectal cancer screening. By analyzing gut microbiota in stool samples with machine learning, they achieve a 90% detection rate, nearly equal to colonoscopies at 94%. This method surpasses all other non-invasive tests, offering a simpler, more accessible early screening option to improve treatment success and save lives. The team created a detailed catalogue of human gut microbiota subspecies and combined it with clinical data to accurately predict the presence of colorectal cancer. Their innovative approach to use machine learning for analyzing microbiota at the subspecies level identifies minute bacterial differences indicative of cancer. This breakthrough not only makes colorectal cancer screening less invasive and more available but also sets the stage for developing similar diagnostic tools for various diseases based on gut microbiota, marking a significant advance in medical diagnostics.