Comparison of Deep Learning Models for Diabetes Prediction

Abstract

Diabetes mellitus is a chronic condition affecting millions globally. Early and accurate prediction of diabetes is crucial for timely intervention and effective management. This study investigates the predictive capabilities of five deep learning models—Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Deep Belief Network (DBN)—using a publicly available dataset. The models were trained and evaluated using different train-test splits and assessed based on accuracy, precision, and recall. Among the models, RNN achieved the highest accuracy and precision, while DBN recorded the highest recall, indicating strong performance in detecting true positives. CNN demonstrated consistent and balanced performance across splits, making it a reliable baseline. This comparative analysis highlights the trade-offs between different deep learning architectures and identifies the most effective approaches for diabetes prediction based on specific evaluation criteria.