Answers are grounded in this paper and cite their sources.
Demonstrates enhanced diabetes prediction accuracy using a fusion model of multiple machine learning techniques, indicating significant potential for healthcare applications.
Introduction: Diabetes has become a critical global public health issue due to its increasing prevalence and associated complications. While various machine learning algorithms have been explored for diabetes prediction and classification, the prediction accuracy remains suboptimal, necessitating more effective approaches. Materials and Methods: This study utilized a publicly available diabetes dataset from Kaggle and proposed a predictive model that integrates Deep Neural Networks (DNN), Support Vector Machines (SVM), and Particle Swarm Optimization (PSO). The DNN architecture extracts key feature vectors using ReLU activation functions in the hidden layers and a Softmax function in the output layer. These extracted features are subsequently passed into the SVM for classification. The PSO algorithm is employed to optimize model parameters by systematically exploring the SVM parameter space through an iterative particle swarm updating mechanism, with the mean F1-score from 5-fold cross-validation used as the evaluation metric to enhance classification performance. Results: Experimental evaluation revealed that the proposed fusion model achieved a diabetes prediction accuracy of 96.5%, outperforming the individual models in classification performance. Discussion: The integration of DNN, SVM, and PSO enhances both the feature extraction and classification capabilities of the model. The DNN effectively captures nonlinear patterns in the dataset, while the SVM provides robust classification, and PSO fine-tunes model parameters for optimal results. Conclusion: The fusion model demonstrates significant improvements in diabetes prediction accuracy, indicating its potential applicability in real-world healthcare prediction systems and its effectiveness over standalone models.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: