Boruta Feature Selection and Deep Learning for Alzheimer’s Disease Classification

Alzheimer’s Disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline, memory impairment, and functional deterioration. The early and accurate classification of AD is crucial for timely intervention and management. This study utilizes the Boruta feature selection method to identify the most relevant features for AD classification, selecting the top 15 features based on importance ranking. Three machine learning models—Deep Neural Networks (DNN), Long Short-Term Memory Networks (LSTM), and Support Vector Machines (SVM)—were evaluated using accuracy, precision, recall, and F1-score as performance metrics. The LSTM model demonstrated the highest accuracy (89.30%), outperforming DNN (88.14%) and SVM (84.19%), owing to its capability of capturing temporal dependencies in inpatient data. Results indicate that deep learning models offer superior performance compared to traditional machine learning approaches in AD classification. The study emphasizes the importance of cognitive, lifestyle, and metabolic features in AD diagnosis while acknowledging limitations such as dataset constraints and model interpretability. Future research should improve explainability, incorporate multi-modal data, and leverage real-time monitoring techniques for enhanced AD detection.