Predicting the Role of the Human Gut Microbiome in Constipation Using Machine-Learning Methods: A Meta-Analysis

Abstract

(1) Background: Constipation is a common condition that affects the health and the quality of life of patients. Recent studies have suggested that the gut microbiome is associated with constipation, but these studies were mainly focused on a single research cohort. Thus, we aimed to construct a classification model based on fecal bacterial and identify the potential gut microbes’ biomarkers. (2) Methods: We collected 3056 fecal amplicon sequence data from five research cohorts. The data were subjected to a series of analyses, including alpha- and beta-diversity analyses, phylogenetic profiling analyses, and systematic machine learning to obtain a comprehensive understanding of the association between constipation and the gut microbiome. (3) Results: The alpha diversity of the bacterial community composition was higher in patients with constipation. Beta diversity analysis evidenced significant partitions between the two groups on the base of gut microbiota composition. Further, machine learning based on feature selection was performed to evaluate the utility of the gut microbiome as the potential biomarker for constipation. The Gradient Boosted Regression Trees after chi2 feature selection was the best model, exhibiting a validation performance of 70.7%. (4) Conclusions: We constructed an accurate constipation discriminant model and identified 15 key genera, including Serratia, Dorea, and Aeromonas, as possible biomarkers for constipation.