Scholarlink

<p>Parkinson’s disease (PD) poses a detection challenge due to its concealed nature and the long-term data collection for subjects. This study proposes a PD detection method based on the artificial bee colony and support vector machine (ABC-SVM), which can select the optimal channels with short-time resting-state electroencephalography (EEG) data. Power spectral density (PSD) and differential entropy (DE) are extracted from the data, followed by an evaluation of various time intervals and channels to ascertain the most effective ones. Subsequently, ABC-SVM is proposed to select the optimal channels and get the detection results. The University of New Mexico (UNM) dataset and University of Iowa (UI) dataset are employed in the paper; the results demonstrate that the average accuracy achieved by the two datasets based on their optimal channels is 99.96% and 93.99%. Overall, we use about half of raw data to get better detection results and improve subject comfort significantly.</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

Optimal channels selection based on ABC-SVM in Parkinson′s disease detection using short-time resting state EEG

  • 期刊论文

  • 发表日期：

    2025年6月6日

查看全文
  • Xiaodan Zhang,
  • Lu Zhang,
  • Kemeng Xu,
  • Yuyu Yang,
  • Rui Zhao,
  • Yichong She

摘要

Parkinson’s disease (PD) poses a detection challenge due to its concealed nature and the long-term data collection for subjects. This study proposes a PD detection method based on the artificial bee colony and support vector machine (ABC-SVM), which can select the optimal channels with short-time resting-state electroencephalography (EEG) data. Power spectral density (PSD) and differential entropy (DE) are extracted from the data, followed by an evaluation of various time intervals and channels to ascertain the most effective ones. Subsequently, ABC-SVM is proposed to select the optimal channels and get the detection results. The University of New Mexico (UNM) dataset and University of Iowa (UI) dataset are employed in the paper; the results demonstrate that the average accuracy achieved by the two datasets based on their optimal channels is 99.96% and 93.99%. Overall, we use about half of raw data to get better detection results and improve subject comfort significantly.