Scholarlink

<p>Feature selection is a crucial process in disease detection, especially when analyzing Electroencephalography (EEG) signals, as it significantly impacts classification performance. Inefficient feature selection can degrade the accuracy of classifiers, leading to unreliable disease detection. In this research, a novel optimal feature selection procedure based on Balanced Adaptive Deer Hunting Optimization (BADHO) is proposed to extract optimal features from EEG signals for Alzheimer’s disease and epilepsy detection. The extracted features are classified using a fine-tuned deep learning-based classifier, integrating Convolutional Bidirectional Long Short-Term Memory (CBiLSTM) and Enhanced Grey Wolf Optimization (EGWO) for parameter optimization. This hybrid approach enhances classification accuracy by selecting discriminative features while reducing computational complexity. The proposed GOCBiLSTM model is evaluated using benchmark Bonn and Dementia EEG datasets. Experimental results demonstrate that the model achieves 95.37% accuracy for the Bonn dataset and 97.50% accuracy for the Dementia dataset, significantly outperforming traditional machine learning and existing deep learning models. The high precision, recall, and F1-score values further validate the robustness of the proposed framework. These findings confirm that the optimized feature selection and deep learning-based classification approach provides an efficient and accurate automated disease detection system for real-world EEG based healthcare applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Novel feature selection and optimized classifier for disease detection from EEG signals using hybrid deep learning algorithm

  • 期刊论文

  • 发表日期：

    2025年6月18日

查看全文
  • P Jasphin Jeni Sharmila,
  • T S Shiny Angel

摘要

Feature selection is a crucial process in disease detection, especially when analyzing Electroencephalography (EEG) signals, as it significantly impacts classification performance. Inefficient feature selection can degrade the accuracy of classifiers, leading to unreliable disease detection. In this research, a novel optimal feature selection procedure based on Balanced Adaptive Deer Hunting Optimization (BADHO) is proposed to extract optimal features from EEG signals for Alzheimer’s disease and epilepsy detection. The extracted features are classified using a fine-tuned deep learning-based classifier, integrating Convolutional Bidirectional Long Short-Term Memory (CBiLSTM) and Enhanced Grey Wolf Optimization (EGWO) for parameter optimization. This hybrid approach enhances classification accuracy by selecting discriminative features while reducing computational complexity. The proposed GOCBiLSTM model is evaluated using benchmark Bonn and Dementia EEG datasets. Experimental results demonstrate that the model achieves 95.37% accuracy for the Bonn dataset and 97.50% accuracy for the Dementia dataset, significantly outperforming traditional machine learning and existing deep learning models. The high precision, recall, and F1-score values further validate the robustness of the proposed framework. These findings confirm that the optimized feature selection and deep learning-based classification approach provides an efficient and accurate automated disease detection system for real-world EEG based healthcare applications.