Alterations in Gut Microbiota-Brain Axis in Major Depressive Disorder as Identified by Machine Learning

Abstract

Major depressive disorder (MDD) is a complex mental health condition whose causes may extend beyond purely biological explanations and are increasingly understood within wider ecological and social frameworks. Emerging research on the human gut-brain axis with the help of statistical and artificial intelligence tools aims to elucidate the links between the gut microbiota, diet, environment, and MDD. In this study, we analyzed data from the American Gut Project (AGP), including 361 control and 23 MDD samples, to find potential biomarkers associated with MDD. While alpha and beta diversity analyses revealed no significant differences except for age, multiple differential abundance tools and machine learning (ML) models (Random Forest and XGBoost), whose results were analyzed using Shapley Additive Explanations values, consistently detected a decrease in Bifidobacterium adolescentis and increases in Odoribacter, Ruminococcus, and Adlercreutzia among MDD samples. These four organisms influence inflammation, neurotransmitter balance, gut permeability, and other pathways associated with depression and thus can be recognized as potential biomarkers for MDD. This study highlights the promise of ML to decode the gut-brain axis as a first step in biomarker discovery, thus providing new possibilities for a personalized treatment approach and an improvement in diagnostic tools for MDD.

Keywords: Random Forest; biomarkers; gut microbiome; machine learning; major depressive disorder; microbiota; personalized medicine.