Survey on Early Diabetes Prediction Using AI Models and EHR Data

Diabetes is a chronic disease affecting millions worldwide, leading to severe complications if notdetected early. Early prediction enables timely medical intervention and lifestyle changes, reducing healthrisks. Machine learning (ML) techniques have proven effective in predicting diabetes using clinical andlifestyle-related data. Among them, the Light Gradient Boosting Machine (LightGBM) is highly efficient inhandling large datasets, offering faster training speed and improved accuracy. Compared to traditionalmodels like logistic regression and decision trees, LightGBM excels in feature selection, reduces overfitting,and enhances predictive performance. This survey explores various ML techniques for early diabetesprediction, focusing on LightGBM’s advantages. The integration of ML in healthcare can improve earlydiagnosis, optimize treatment plans, and enhance patient outcomes, making LightGBM a valuable tool indiabetes management.