A framework for the development of an optimized artificial intelligence model for diabetes mellitus prediction and treatment recommendation

Diabetes mellitus presents a significant public health challenge, especially in Malaysia, with its rapidly increasing prevalence. By 2024, an estimated 20.1% of the Malaysian population will be affected by diabetes, potentially rising to 5 million by 2030. This study presents a framework that leverages the Indian Pima Diabetes dataset to develop and evaluate an optimized artificial intelligence (AI) model for diabetes prediction and treatment guidance. Combining machine learning, deep learning algorithms, and ensemble techniques like model stacking, the framework aims to achieve high prediction accuracy, balancing sensitivity and specificity, to support clinical decision-making. The study also highlights the importance of addressing ethical considerations, data privacy, and algorithmic biases to harness AI's full potential in transforming diabetes care.