Scholarlink

<p>Epilepsy is a serious neurological disorder that remains difficult to detect with high accuracy. Automated seizure detection using EEG signals has gained increasing attention, and optimization algorithms are often applied to improve system performance. Atom Search Optimization (ASO) has strong global search ability but frequently suffers from premature convergence and limited local search efficiency. To address these issues, this study proposes a hybrid algorithm that combines ASO with the Sine–Cosine Algorithm (SCA) and a ranking strategy (RSHASOSCA). ASO provides effective global exploration, SCA enhances local exploitation, and the ranking strategy stabilizes convergence, together creating a more balanced and reliable search process. The method was evaluated on the CHB-MIT scalp EEG dataset. Features were extracted using Wavelet Packet Transform (WPT) and refined with the Kruskal–Wallis test (<i>p</i> ≤ 0.001). Comparative experiments against twelve established optimization algorithms showed that the RSHASOSCA framework achieved superior performance. When applied with an SVM classifier, it reached 99.13% accuracy and an AUC of 1. These findings highlight the value of integrating ASO, SCA, and ranking strategy, and demonstrate the potential of the proposed framework for reliable and efficient seizure detection in clinical practice.</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

An improved atom search optimization algorithm based on ranking strategy and sine cosine algorithm for epileptic seizure detection

  • 期刊论文

  • 发表日期：

    2025年10月25日

查看全文
  • Kaliprasanna Swain,
  • Tan Kuan Tak,
  • Kamal Upreti,
  • Pravin R. Kshirsagar,
  • Sivaneasan Bala Krishnan,
  • Ramesh Chandra Poonia,
  • Sumant Kumar Mohapatra,
  • Soumya Ranjan Nayak

摘要

Epilepsy is a serious neurological disorder that remains difficult to detect with high accuracy. Automated seizure detection using EEG signals has gained increasing attention, and optimization algorithms are often applied to improve system performance. Atom Search Optimization (ASO) has strong global search ability but frequently suffers from premature convergence and limited local search efficiency. To address these issues, this study proposes a hybrid algorithm that combines ASO with the Sine–Cosine Algorithm (SCA) and a ranking strategy (RSHASOSCA). ASO provides effective global exploration, SCA enhances local exploitation, and the ranking strategy stabilizes convergence, together creating a more balanced and reliable search process. The method was evaluated on the CHB-MIT scalp EEG dataset. Features were extracted using Wavelet Packet Transform (WPT) and refined with the Kruskal–Wallis test (p ≤ 0.001). Comparative experiments against twelve established optimization algorithms showed that the RSHASOSCA framework achieved superior performance. When applied with an SVM classifier, it reached 99.13% accuracy and an AUC of 1. These findings highlight the value of integrating ASO, SCA, and ranking strategy, and demonstrate the potential of the proposed framework for reliable and efficient seizure detection in clinical practice.