Scholarlink

This study investigates EEG-based biomarkers and machine learning for objective assessment of motor senescence interventions in elderly populations. We recorded resting-state and movement-related EEG signals from participants aged 40+ before and after therapy, extracting power spectral density (PSD) features to characterize age-related neural patterns. Six machine learning models (LR, DT, RF, kNN, SVM, XGB) were evaluated for age-group discrimination. XGB achieved peak performance (90.4% accuracy during rest; 90.2% during movement), with Random Forest closely following (88.1%; 86.7%). Spectral analysis revealed elevated \(\alpha \) / \(\beta \) -band PSD in elderly participants, suggesting compensatory neural mechanisms, while movement preparation showed consistent \(\alpha \) -band suppression across age groups. Classification metrics demonstrated superior recognition of young adults (precision: 92–93% vs. 87% for elderly), highlighting a systematic model bias. These findings validate EEG biomarkers as quantitative tools for assessing neuroplasticity-driven motor recovery, addressing critical gaps in geriatric care evaluation through computational neurophysiology.

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

EEG-Based Discrimination of Aging Effects from Movement Therapy Using Machine Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Luis G. Hernandez-Rojas,
  • Amanda Valdez-Calderon,
  • David Cruz-Ortiz,
  • Jorge German Perez Blanco,
  • Sofia C. Henao,
  • Mariana Felisa Ballesteros-Escamilla,
  • Joel C. Huegel,
  • Javier M. Antelis

摘要

This study investigates EEG-based biomarkers and machine learning for objective assessment of motor senescence interventions in elderly populations. We recorded resting-state and movement-related EEG signals from participants aged 40+ before and after therapy, extracting power spectral density (PSD) features to characterize age-related neural patterns. Six machine learning models (LR, DT, RF, kNN, SVM, XGB) were evaluated for age-group discrimination. XGB achieved peak performance (90.4% accuracy during rest; 90.2% during movement), with Random Forest closely following (88.1%; 86.7%). Spectral analysis revealed elevated \(\alpha \) / \(\beta \) -band PSD in elderly participants, suggesting compensatory neural mechanisms, while movement preparation showed consistent \(\alpha \) -band suppression across age groups. Classification metrics demonstrated superior recognition of young adults (precision: 92–93% vs. 87% for elderly), highlighting a systematic model bias. These findings validate EEG biomarkers as quantitative tools for assessing neuroplasticity-driven motor recovery, addressing critical gaps in geriatric care evaluation through computational neurophysiology.