Machine learning approach reveals microbiome, metabolome, and lipidome profiles in type 1 diabetes.

INTRODUCTION

Type 1 diabetes (T1D) is a complex disorder influenced by genetic and environmental factors. The gut microbiome, the serum metabolome, and the serum lipidome have been identified as key environmental factors contributing to the pathophysiological mechanisms of T1D.

OBJECTIVES

We aimed to explore the gut microbiota, serum metabolite, and serum lipid signatures in T1D patients by machine learning.

METHODS

We evaluated 137 individuals in a cross-sectional cohort involving 38 T1D patients, 38 healthy controls, and 61 T1D patients for validation. We characterized gut microbiome, serum metabolite, and serum lipid profiles with machine learning approaches (logistic regression, support vector machine, Gaussian naive Bayes, and random forest).

RESULTS

The machine learning approaches using the microbiota composition did not accurately diagnose T1D (model accuracy = 0.7555), while the accuracy of the model using the metabolite composition was 0.9333. Based on the metabolite composition, 3-hydroxybutyric acid and 9-oxo-ode (area under curve = 0.70 and 0.67, respectively, both increased in T1D) were meaningful overlap metabolites screened by multiple bioinformatics methods. We confirmed the biological relevance of the microbiome, metabolome, and lipidome features in the validation group.

CONCLUSION

By using machine learning algorithms and multi-omics, we demonstrated that T1D patients are associated with altered microbiota, metabolite, and lipidomic signatures or functions.