Scholarlink

<b>Purpose:</b> <p>This study aimed to develop an efficient and practical approach for insomnia detection using single-channel electrooculogram (EOG) signals analyzed through multiscale entropy (MSE).</p> <b>Methods:</b> <p>EOG recordings from 639 participants (257 healthy controls and 382 patients with insomnia) were used to construct a classification framework based on Gradient Boosting of MSE-derived features. Analyses were stratified by age groups (0–29, 30–44, 45–64, and <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\ge\)</EquationSource> </InlineEquation>65 years). Only the first 80 minutes of sleep data were required to train and evaluate models.</p> <b>Results:</b> <p>Age-stratified models yielded substantially improved performance compared with the non-stratified approach. MSE-based classifiers achieved accuracies of 86.4%, 87.6%, 84.2%, and 91.2% for respective age groups, outperforming the 80.2% accuracy achieved without stratification. MSE and RCMSE values were consistently lower in insomnia patients, reflecting reduced signal complexity compared to healthy controls. The optimal recording duration ranged between 55 and 80 minutes across age groups.</p> <b>Conclusions:</b> <p>Short, age-tailored EOG recordings analyzed with MSE provide an accessible and reliable method for preliminary insomnia screening. The method’s simplicity and ability to function without sleep staging make it a practical alternative to resource-intensive polysomnography, suitable for integration into wearable devices for home-based screening.</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

Age-specific insomnia detection models using short-time EOG and multiscale entropy analysis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月23日

查看全文
  • Chih-Lin Lin,
  • Min-Huey Chung,
  • Ching-I Huang,
  • Chih-En Kuo

摘要

Purpose:

This study aimed to develop an efficient and practical approach for insomnia detection using single-channel electrooculogram (EOG) signals analyzed through multiscale entropy (MSE).

Methods:

EOG recordings from 639 participants (257 healthy controls and 382 patients with insomnia) were used to construct a classification framework based on Gradient Boosting of MSE-derived features. Analyses were stratified by age groups (0–29, 30–44, 45–64, and \(\ge\) 65 years). Only the first 80 minutes of sleep data were required to train and evaluate models.

Results:

Age-stratified models yielded substantially improved performance compared with the non-stratified approach. MSE-based classifiers achieved accuracies of 86.4%, 87.6%, 84.2%, and 91.2% for respective age groups, outperforming the 80.2% accuracy achieved without stratification. MSE and RCMSE values were consistently lower in insomnia patients, reflecting reduced signal complexity compared to healthy controls. The optimal recording duration ranged between 55 and 80 minutes across age groups.

Conclusions:

Short, age-tailored EOG recordings analyzed with MSE provide an accessible and reliable method for preliminary insomnia screening. The method’s simplicity and ability to function without sleep staging make it a practical alternative to resource-intensive polysomnography, suitable for integration into wearable devices for home-based screening.