Computational approaches for clinical, genomic and proteomic markers of response to glucagon-like peptide-1 therapy in type-2 diabetes mellitus: An exploratory analysis with machine learning algorithms

Introduction: In 2021, the International Diabetes Federation reported that 537 million people worldwide are living with diabetes. While glucagon-like peptide-1 agonists provide significant benefits in diabetes management, approximately 40 % of patients do not respond well to this therapy. This study aims to enhance treatment outcomes by using machine learning to predict individual response status to glucagon-like peptide-1 therapy. Methods: We analysed a type-2 diabetes mellitus dataset from the Diastrat cohort, recruited at the Northern Ireland Centre for Stratified Medicine. The dataset included individuals prescribed glucagon-like peptide-1 therapy, with response status determined by glycated haemoglobin levels of <= 53 mmol/mol. We identified genomic and proteomic markers and developed machine learning models to predict therapy response. Results: The study found 5 genomic variants and 45 proteomic markers that help differentiate glucagon-like peptide-1 therapy responders from non-responders, achieving 95 % prediction accuracy with a machine learning model. Conclusion: This study demonstrates the potential of machine learning in predicting the response to glucagon-like peptide-1 therapy in individuals with type-2 diabetes mellitus. These findings suggest that integrating genomic and proteomic data can significantly enhance personalized treatment approaches, potentially improving outcomes for patients who might otherwise not respond well to glucagon-like peptide-1 therapy. Further research and validation in larger cohorts are necessary to confirm these results and translate them into clinical practice.