Scholarlink

<p>Acute encephalopathy with biphasic seizures and late reduced diffusion (AESD) is the most common subtype of acute encephalopathy in Japan and is difficult to differentiate from prolonged febrile seizures (PFSs). This study aimed to explore the capability of machine learning to differentiate AESD from PFSs on the basis of early electroencephalogram (EEG) analyses. Sixty one children with AESD (<i>n</i> = 20) or PFS (<i>n</i> = 41) were included. Digital EEG data with bipolar montage collected within 48&#xa0;h (1–48&#xa0;h) after seizure onset were analyzed using absolute power spectrum (APS) and phase lag index (PLI) values in each EEG frequency band. The APS values in the theta, alpha, beta, and gamma bands were lower for AESD than those for PFS. By contrast, the mean PLI values for all frequency bands were higher for AESD than for PFS. Machine learning analysis revealed that the APS value in the beta bands provided the highest differentiation accuracy and positive predictive value for AESD (68.8%). The mean APS values across all electrodes in the beta band may be a useful tool for differentiating between early-phase AESD and PFS. This study demonstrates the potential for early automated diagnosis of AESD and PFS using EEG analysis.</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

Automated differentiation of acute encephalopathy with biphasic seizures and late reduced diffusion and prolonged febrile seizures in acute phase

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月26日

查看全文
  • Masayoshi Oguri,
  • Tohru Okanishi,
  • Ichiro Kuki,
  • Shin-Ichiro Hamano,
  • Ikuya Ueta,
  • Yuko Nakamura,
  • Yuji Inaba,
  • Sooyoung Lee,
  • Jun-Ichi Takanashi,
  • Masami Togawa,
  • Akinari Onishi,
  • Yoshihiro Maegaki

摘要

Acute encephalopathy with biphasic seizures and late reduced diffusion (AESD) is the most common subtype of acute encephalopathy in Japan and is difficult to differentiate from prolonged febrile seizures (PFSs). This study aimed to explore the capability of machine learning to differentiate AESD from PFSs on the basis of early electroencephalogram (EEG) analyses. Sixty one children with AESD (n = 20) or PFS (n = 41) were included. Digital EEG data with bipolar montage collected within 48 h (1–48 h) after seizure onset were analyzed using absolute power spectrum (APS) and phase lag index (PLI) values in each EEG frequency band. The APS values in the theta, alpha, beta, and gamma bands were lower for AESD than those for PFS. By contrast, the mean PLI values for all frequency bands were higher for AESD than for PFS. Machine learning analysis revealed that the APS value in the beta bands provided the highest differentiation accuracy and positive predictive value for AESD (68.8%). The mean APS values across all electrodes in the beta band may be a useful tool for differentiating between early-phase AESD and PFS. This study demonstrates the potential for early automated diagnosis of AESD and PFS using EEG analysis.