Machine Learning Approach to Identify Gut Microbiota Biomarkers in Patients with ST-Elevation Myocardial Infarction Presenting Primary Ventricular Tachyarrhythmias
Journal:
medRxiv
Published Date:
Sep 20, 2026
Abstract
Background Risk stratification for primary ventricular tachycardia/ventricular fibrillation (VT/VF) in patients with ST-elevation myocardial infarction (STEMI) remains limited. We aimed to identify gut microbiome biomarkers and microbial metabolic pathways associated with primary VT/VF in STEMI using machine learning (ML). Methods Stool samples from 33 STEMI patients (7 with primary VT/VF and 26 without VT/VF) underwent full-length 16S rRNA sequencing during the acute ([≤]7 days) and recovery (30?60 days) phases. An ML pipeline integrating taxonomic features identified reproducible features associated with primary VT/VF. Core biomarkers were defined as taxa demonstrating both univariate significance (false discovery rate [FDR] <0.05) and multivariate classification importance. Functional pathways and enzymes were predicted using PICRUSt2. Results Five acute-phase taxa were reproducibly associated with primary VT/VF: Clostridium aldenense, Enterocloster bolteae, bacterium NLAE-zl-G101, Alistipes shahii, and Roseburia sp. (all FDR <0.05). A reduced-feature support vector machine model achieved an area under the receiver operating characteristic curve of 0.846 (95% confidence interval, 0.676?0.984), with 85.7% sensitivity and 76.9% specificity. Differential abundance analysis showed enrichment of Bacteroides fragilis, Bacteroides thetaiotaomicron, and E. bolteae. Functional analysis demonstrated enrichment of fucose degradation and purine catabolism, with increased predicted abundances of L-fucose mutarotase and xanthine dehydrogenase. Conclusions Primary VT/VF in STEMI was associated with a distinct gut microbiome-metabolic signature characterized by Lachnoclostridium-related taxa, persistent Bacteroides enrichment, and enhanced predicted fucose and purine metabolism. These exploratory findings warrant validation in larger independent cohorts and mechanistic studies.
