Diabetes Estimation Through Data Mining Using Optimization, Clustering, and Secure Cloud Storage Strategies

Asian Indians face a unique risk for type 2 diabetes, with earlier onset and diagnosis at lower BMIs than other populations. Their susceptibility to diabetes-related health risks is higher compared to white populations. India struggles with low public awareness, a shortage of healthcare professionals, and limited treatment accessibility. Innovative strategies using available technological resources could significantly improve diabetes management in India. These challenges, common in developing countries, suggest that insights from India’s experience could benefit global diabetes management. This study aims to develop a hybrid model for predicting type 2 diabetes mellitus (T2-DM) to enhance awareness and preventive strategies. Using the Pima Indian Diabetes Dataset, the model’s effectiveness is validated through various performance metrics. The Fuzzy C-Means (FCM) clustering method is introduced, outperforming traditional models by creating more cohesive clusters with fewer variables. To enhance FCM's effectiveness, the Particle Swarm Optimization (PSO) technique is integrated, preserving FCM's strengths while improving efficiency. The hybrid model's sensitivity is 5.85% higher than K-Means + C4.5, and specificity is 2.79% higher. The proposed model achieves greater accuracy (96.58%) than previous methods. This approach aims to improve diabetes care in India and offers a model adaptable worldwide, addressing a critical global health challenge.