OptiANN-LR: Augmenting Diabetes Prediction Accuracy through Hyper Learning Rate Tuning in Optimized Artificial Neural Networks

• Main Authors: P, Sujitha S, Anita, C. S., Sudharson, K., Rose, S. Remya, S, Sharmila, V, Keerthana
• Format: Conference Proceeding
• Language: English
• Subjects: Artificial neural networks; Diabetes; diabetes prediction; early diagnosis; hyperparameter tuning; Learning (artificial intelligence); machine learning in healthcare; Medical services; Parallel processing; Software; Training
• ISSN: 2688-0288
• DOI: 10.1109/SCEECS61402.2024.10481943

Diabetes, a pervasive chronic condition, presents global health challenges. We propose a novel approach to address this, optimizing Artificial Neural Networks (ANNs) through hyperparameter tuning. Our model, OptiANN-LR, achieves an exceptional 95.57% accuracy, surpassing ABP-SCGNN 93%. Success hinges on meticulous learning rate tuning, a crucial ANN hyperparameter. With the Adam optimizer, OptiANN-LR efficiently converges, and dropout layers curb overfitting, ensuring robust generalization. These enhancements promise early diabetes diagnosis and prevention. Rigorous evaluations employing k-fold cross-validation and diverse metrics affirm OptiANN-LR's reliability. Our work showcases advanced machine learning in healthcare, especially diabetes prediction, advancing data-driven approaches. This research heralds more accurate healthcare solutions, potentially benefiting millions at risk of diabetes and facilitating proactive interventions.