Deciphering Post-surgery Gut Microbial Dynamics in Colorectal Cancer through Multi-cohort Machine Learning

Surgical resection, and its associated bowel preparation, remain the primary treatment for colorectal cancer (CRC), yet the associated effects on postoperative gut microbiota composition remain poorly characterized. This study profiled gut microbial compositional changes before and after colorectal surgery and identified microbial signatures associated with postoperative recovery. Four external study cohorts were integrated to strengthen biological interpretation of our internal cohort observations. Leveraging Machine learning (ML) combined with SHapley Additive exPlanations (SHAP), we identified specific taxa contributing to pre- and post-operative stratification such as Akkermansia, Bacteroides and Streptococcus. Specifically, Akkermansia was identified as a major contributor to the post-operative microbiota stratification, supporting its potential as a microbial biomarker for post- operative recovery. Our study lays the groundwork for future studies on the key identified microbial signatures for postoperative recovery monitoring and microbiota-targeted therapeutic strategies.