T2D is a world-wide-associated disease which creates large social and economic costs. Annex IV Early Detection access and intervention Application; there are several complications associated with diabetes; hence, early detection ensures the patient gets early treatment leading to improved outcomes. In the present study, we introduce a prediction model aimed at using AI/ML and large datasets to predict patients who are at increased risk of developing T2D. It incorporates EHRs, genomics data, lifestyle information, and socioeconomic status as a diverse set of risk factors the risk profile was then extracted by applying Random Forest, Gradient Boosting Machines and Deep Learning models and applied on large data sets for modelling. Thus, to enhance the quality of data used, some data preprocessing methods such as normalization, method of outlier, and missing value imputation were conducted. In the present study, feature selection techniques like SHAP (Shapley Additive explanations) were applied to determine the significance of potential risk factors. Meaningful performance of the models implemented in this study indicates that the system has high accuracy and precision in detecting pre-diabetic patients. Moreover, the integration of big data analytics that differ from traditional techniques made it possible to process data in real-time and scale up the model to be easily implemented in clinical practices and programs in the field of public health. This research establishes the applicability of data-driven AI, particularly within ML frameworks, to assist in the early diagnosis of T2D as well as the development of individualistic healthcare solutions. The proposed framework has the potential of facilitating targeted preventive measures by presenting an actionable risk prediction model to caregivers and policy makers to fight T2D hence rationalizing the overall disease burden.