Diabetes is a chronic condition, and with the increasing prevalence of cases globally, there is a heightened need for innovative IT solutions that facilitate early detection and effective management to minimize its impact on individuals and healthcare systems. This research evaluates the performance of two machine learning models–Random Forest and XGBoost in predicting the onset of diabetes. The Random Forest algorithm achieved an impressive 95% accuracy, outperforming XGBoost, which reached 91%. Moreover, by utilizing the Apache Spark framework, we reduced the model’s training time to just 5–10 seconds, a considerable improvement compared to the traditional Random Forest model, which required 16 min. To put these results into practical use, we developed a web application utilizing the Apache Spark Random Forest model to assess diabetes risk in new cases, offering personalized health insights. This study highlights the potential of machine learning for early diabetes prediction, demonstrating the superiority of the Random Forest model and suggesting areas for improving XGBoost. Overall, it showcases how predictive modeling can be leveraged for managing chronic conditions and advancing personalized healthcare solutions for diabetes.