Machine Learning-based Investigation Of The Relationship Between Gut Microbiome And Obesity Status

The human gut microbiome has about 150 times more genes than the human body,and it is a complex ecosystem.It is not only affected by the host’s survival status,but it also has a profound impact on the host’s health and disease.In the past 10 years,a large number of studies have found that the gut microbiome is closely related to obesity.However,in the published research results on the interaction between the gut microbiome and obesity,there are few consistent conclusions,and most of the methods and results are not generalizable.Among them,the small scale of the study cohort and the single study method are important reasons for this phenomenon.In order to determine the difference in the gut microbiome between normal individuals and obesity individuals,our study collected and analyzed the stool samples and clinical phenotype information of 2262 Chinese people.First,this study used bioinformatics methods to calculate the relative abundance,αdiversity,βdiversity,and distance matrix of gut microbiome species and metabolic pathways,and statistical methods Mann-Whitney U test and PCo A analysis were performed to compare the differences in gut microbiome diversity of individuals with different obesity status.Then,RF,SVM,KNN,LR,LASSO,GBDT and RR algorithms were used to build regression and classification models to predict obesity status through gut microbiome information.In addition,this study combined the Lefse method with an iterative method based on machine learning to explore the characteristics of gut microbiome that can affect obesity.The results of the study showed that compared with normal individuals,obesity individuals showed lower species diversity and higher metabolic pathway diversity.In addition,various machine learning models were used to quantify the relationship between obesity status and the gut microbiome.Among them,the support vector machine models achieved the best performance with a classification accuracy of 0.716 and a regression R~2 score of 0.485.In addition to two species that have been shown to be associated with obesity,we also identified three new species that have the potential to become obesity-related biomarkers,including Bacteroides caccae,Odoribacter splanchnicus and Roseburia hominis.Further analysis of metabolic pathways also revealed that metabolic pathways such as lipid biosynthesis(lipid IVA biosynthesis)and panthenol biosynthesis are significantly enriched in the gut of obesity individuals.In summary,the results of our study showed that there was a close connection between obesity and the gut microbiome of a large Chinese population.The machine learning method in the study had universal applicability to the study of the gut microbiome of the Chinese population.The identified species and metabolic pathways that have the potential to become obesity-related biomarkers may provide new ideas and targets for the prevention and treatment of obesity.