Development and External Validation of Machine Learning Models for Diabetic Microvascular Complications: Cross-Sectional Study With Metabolites

Abstract

This study highlighted diabetes duration, insulin usage, age, and circulating tyrosine as important factors in detecting DKD and DR. The integration of ML with biomedical big data enables biomarker discovery and improves disease detection beyond traditional risk factors.