Scholarlink

<p>Internet of Medical Things (IoMT) has changed healthcare through cost reduction, enhanced accessibility, and operational effectiveness. Early diagnosis of neurological brain disorders remains a significant issue despite all of these advancements because of the tremendous complexity and variability of EEG data. Spurred by the requirement for reliable, real-time diagnosis in IoMT systems, this study formulates the Dense Nested Dynamic Graph Convolutional Network with Dollmaker Optimization Algorithm (DNDGCNet-DOA). The system pre-processes EEG signals using an Adaptive Self-Guided Loop Filter (AS-GLF), extracts dense features using the Short-Time Quaternion Quadratic Phase Fourier Transform (ST-QQPFT), and chooses optimal subsets based on a Billiards-Inspired Optimization Algorithm (BOA). The advanced features are then passed into the DNDGCNet model, which learns intricate spatial–temporal relations, while network weights are optimized using the Dollmaker Optimization Algorithm (DOA) for better performance. The model produced yields exceptional performance with accuracy rates higher than 99% in both neurotypical vs. epilepsy and autism spectrum disorder classification, as well as comparable levels of precision, recall, and AUC scores. The model provides a strong, interpretable solution, opening the door to real-time clinical implementation and general neurological disorder diagnosis.</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

Real-Time EEG-Driven Diagnostic Decision Support: Dense Nested Dynamic Graph Convolutional Network Model for Neurological Disorder Detection

  • 期刊论文

  • 发表日期：

    2025年10月19日

查看全文
  • Jampani Chandra Sekhar,
  • Jarubula Ramu,
  • M. Senthil Vadivu,
  • Krishna Prakash Arunachalam

摘要

Internet of Medical Things (IoMT) has changed healthcare through cost reduction, enhanced accessibility, and operational effectiveness. Early diagnosis of neurological brain disorders remains a significant issue despite all of these advancements because of the tremendous complexity and variability of EEG data. Spurred by the requirement for reliable, real-time diagnosis in IoMT systems, this study formulates the Dense Nested Dynamic Graph Convolutional Network with Dollmaker Optimization Algorithm (DNDGCNet-DOA). The system pre-processes EEG signals using an Adaptive Self-Guided Loop Filter (AS-GLF), extracts dense features using the Short-Time Quaternion Quadratic Phase Fourier Transform (ST-QQPFT), and chooses optimal subsets based on a Billiards-Inspired Optimization Algorithm (BOA). The advanced features are then passed into the DNDGCNet model, which learns intricate spatial–temporal relations, while network weights are optimized using the Dollmaker Optimization Algorithm (DOA) for better performance. The model produced yields exceptional performance with accuracy rates higher than 99% in both neurotypical vs. epilepsy and autism spectrum disorder classification, as well as comparable levels of precision, recall, and AUC scores. The model provides a strong, interpretable solution, opening the door to real-time clinical implementation and general neurological disorder diagnosis.