Colorectal Cancer Gut Microbe Clues Bolstered by AI Approach

In Genome Biology, a team from the Tokyo Institute of Technology, Metagen Therapeutics, and elsewhere outlines colorectal cancer (CRC) clusters marked by distinct gut microbial community associations found with an artificial intelligence (AI) approach. The investigators used a Shapley Additive Explanations (SHAP) algorithm and a SHAP-based microbiome analyses tool known as SHAP-MAT to analyze bacterial abundance profiles from five publicly available datasets representing individuals with or without CRC. In the large datasets and in samples from four individuals with CRC, the authors saw that the SHAP strategy could help to focus in on a range of bacteria-based candidate biomarkers that coincided with distinct CRC clusters. "[W]e explored the potential of using explainable AI for gut microbiome-based CRC classification," they write. "We showed that SHAP could be used to obtain more personalized feature importance that can be used to identify potential bacterial biomarkers for CRC.