Machine Learning-Based Plasma Metabolomics in Liraglutide-Treated Type 2 Diabetes Mellitus Patients and Diet-Induced Obese Mice

Abstract

Liraglutide, a glucagon-like peptide-1 receptor agonist, is effective in the treatment of type 2 diabetes mellitus (T2DM) and obesity. Despite its benefits, including improved glycemic control and weight loss, the common metabolic changes induced by liraglutide and correlations between those in rodents and humans remain unknown. Here, we used advanced machine learning techniques to analyze the plasma metabolomic data in diet-induced obese (DIO) mice and patients with T2DM treated with liraglutide. Among the machine learning models, Support Vector Machine was the most suitable for DIO mice, and Gradient Boosting was the most suitable for patients with T2DM. Through the cross-evaluation of machine learning models, we found that liraglutide promotes metabolic shifts and interspecies correlations in these shifts between DIO mice and patients with T2DM. Our comparative analysis helped identify metabolic correlations influenced by liraglutide between humans and rodents and may guide future therapeutic strategies for T2DM and obesity.