<p>Diabetes Mellitus is a chronic metabolic disorder which is posing severe health problems worldwide because of the complications it is associated with cardiovascular disease, kidney failure, neuropathy, and loss of vision. Conventional diagnosis is based on clinical tests, but the development of Big Data Analytics has made it possible to use data-driven methods of early prediction and control. This paper presents a composite predictive model, which is a combined method of traditional clinical variables (inclusive of glucose, body mass index (BMI), age, and insulin) and additional risk variables (inclusive of lifestyle habits, family history, and blood pressure). This is integrated into an organised pipeline model that includes data preprocessing, machine learning optimisation and feature selection to classify. The experimental results show that introducing extraneous variables played a significant role in the accomplishment of the higher prediction accuracy of 86.3%, in comparison to the lower prediction accuracy of 78.6%, which is a superior model robustness and reliability. Overall, the proposed framework demonstrates that the integration of multi-dimensional risk factors into a structured pipeline could contribute significantly to the process of diabetes prediction in order to deliver a scalable and efficient tool of predictive healthcare analytics and proactive disease management.</p>