A Multi-Layered Learning (MLL) Model for Detecting Diabetes

Diabetes is a chronic metabolic disorder affecting many of the world's population. Diabetes detection is critical for timely intervention and effective disease management. Machine learning techniques, particularly deep learning approaches, have shown promising results in diabetes detection in recent years. The authors propose a multilayered approach for diabetes detection that combines enhanced feature selection techniques with a Fully Connected Recurrent Neural Network (FRNN). The proposed approach's first layer involves feature selection, identifying the most informative features from the input dataset. Traditional feature selection methods, such as correlation analysis and statistical tests, are improved using domain knowledge and data preprocessing techniques. It reduces the dimensionality of the input data and removes irrelevant or redundant features, improving the efficiency and accuracy of the subsequent classification model. FRNN is used in the second layer to detect diabetes. The FRNN is a neural network architecture that effectively captures temporal dependencies in sequential data. It comprises recurrent connections, allowing the network to retain information from previous time steps, making it suitable for analyzing time series data like medical records. The proposed multilayered approach is tested on a large diabetes dataset with various clinical features and patient records. Standard evaluation metrics such as precision, accuracy, recall, specificity, and F1-score used to evaluate the proposed approach's performance. The experimental results show that enhanced feature selection and the FRNN model outperform other diabetes detection methods. The MLL achieves high accuracy and robustness when dealing with the complexity and variability of diabetes data. These findings suggest that a multilayered approach to diabetes detection and management can improve patient outcomes and healthcare efficiency.