A Novel Stacked Ensemble Learning for Early Diabetes Prediction: A Hybrid Approach with 1D-CNN, LSTM, and Traditional Machine Learning Classifiers

Diabetes is a widespread chronic disease that impacts millions of individuals around the globe. Recent advances in artificial intelligence have highlighted the potential of machine learning (ML) and deep learning (DL) techniques for the early detection of diabetes, offering improvements in both diagnostic precision and efficiency. This research introduces an innovative stacked ensemble learning framework that integrates various ML and DL algorithms to boost predictive accuracy. The approach begins by utilizing individual models—such as one-dimensional convolutional neural networks (1D-CNN), long short-term memory (LSTM) networks, random forest (RF), logistic regression (LR), and support vector machines (SVM), each selected for its distinct strengths in prediction. These models are then strategically combined through a stacked ensemble method to capitalize on their complementary capabilities and enhance overall performance. The proposed ensemble model achieves a prediction precision of 0.95, a recall of 0.68, an F1-score of 0.79, and an ROC-AUC of 0.97, demonstrating its effectiveness in early diabetes detection. This integrated approach ensures not only high predictive accuracy but also robustness against variations in patient data, ultimately facilitating timely intervention and improved healthcare outcomes.