Diabetes mellitus, a persistent metabolic disorder, presents a significant global health challenge. Accurate early prediction of diabetes is essential for efficiently managing and averting its associated complications. This study delves into the utilization of Machine Learning (ML) models, encompassing Support Vector Machines (SVM), Logistic Regression (LR), Extra Trees (ET), and Random Forests (RF), for forecasting the emergence of diabetes. To rectify the inherent class imbalance in our dataset, we employ the Synthetic Minority Over-sampling Technique (SMOTE) for equilibrium. The results of our study highlight the Extra Trees (ET) model as the most accurate ML models for predicting diabetes after addressing dataset imbalances, with Random Forests (RF) closely following. These results help us to considerate ML role in health sector and it is showing a significant value to enhance diabetes danger assessment, finally improves the efficiency of health sector intrusions.