A Dual Predictive Autoencoder-Based Model for Diabetes Prediction : Utilizing Supervised Learning and Feature Extraction with an Autoencoder for Accurate Diabetes Prediction

Early prediction of diabetes is crucial for timely diagnosis and treatment. In this study, we propose a Dual Predictive Autoencoder (DP-AE) model for diabetes prediction, aiming to enhance feature learning and classification accuracy. The experimental results demonstrate that the DP-AE model achieves an accuracy of 0.746, precision of 0.739, recall of 0.708, an F1 score of 0.715, and an AUC score of 0.81. Compared to the second-ranked Autoencoder (AE) model, the DP-AE model shows improvements of approximately 3.76% in accuracy, 3.51% in recall, and 3.77% in F1 score. Moreover, the DP-AE model outperforms the second-ranked Multilayer Perceptron (MLP) model by approximately 4.67% in terms of precision and 2.53% in terms of AUC. These findings highlight the potential of the DP-AE model as a valuable tool for diabetes prediction, enabling early detection and intervention for individuals at risk.