Researchers at the University of Geneva have built something the GI field has long been working toward: a stool-based colorectal cancer screening test powered by machine learning that detected 90% of cancer cases — approaching colonoscopy’s 94% detection rate and outperforming every existing non-invasive method. The key innovation wasn’t the use of gut bacteria per se, but the level of precision applied. Rather than analyzing broad bacterial species, the team mapped microbiota at the subspecies level — an intermediate resolution specific enough to capture disease-relevant differences, yet consistent enough to hold across diverse populations.
Keep Reading
Add A Comment
