Optimization Methods for Random Forest Models in Early Diabetes Prediction

Diabetes affects approximately 537 million people worldwide, accounting for 10.5% of the total population. It has a significant impact on individuals' physical and mental health, highlighting the urgent need for low-cost and effective screening methods. This study focuses on using random forest models for early diabetes prediction. We compare and analyze the results using cross-validation, grid search optimization, and polynomial feature engineering methods to calculate the hyperparameters of the random forest model. These optimization techniques effectively improve the prediction accuracy of the random forest model, enabling more accurate identification of early-stage diabetes patients and promoting diabetes prevention, diagnosis, and early treatment.