Scholarlink

<p>The work presents an optimized anomaly identification framework for Wireless Body Area Networks (WBANs), focusing on efficiency and reduced false positives. With constant vital sign monitoring, WBANs are essential to contemporary healthcare. They are intended to be effective and minimally intrusive. In order to discover medical problems early and address them in a timely manner, real-time data collection may be essential. WBANs are increasingly critical for real-time health monitoring, yet the high rate of false-positive alerts and computational constraints impede practical deployment. This work combines an optimized Convolutional Long Short-Term Memory (ConvLSTM) model for capturing temporal and spatial patterns in physiological data with an adaptive Fuzzy Logic-based threshold adjustment to address these challenges. The results obtained with an accuracy of 99.38%, a recall of 99.50%, and an AUC of 0.996 show an acceptable reduction in false- positive rates. The proposed work dominantly reduces the false positives with improved effectiveness for real-time wearable WBAN.</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

Adaptive Fuzzy-Threshold ConvLSTM Model for Efficient Real-Time WBAN Anomaly Identification

  • 期刊论文

  • 发表日期：

    2026年5月14日

查看全文
  • B. Pabitha,
  • V. Vani,
  • Shridhar Sanshi

摘要

The work presents an optimized anomaly identification framework for Wireless Body Area Networks (WBANs), focusing on efficiency and reduced false positives. With constant vital sign monitoring, WBANs are essential to contemporary healthcare. They are intended to be effective and minimally intrusive. In order to discover medical problems early and address them in a timely manner, real-time data collection may be essential. WBANs are increasingly critical for real-time health monitoring, yet the high rate of false-positive alerts and computational constraints impede practical deployment. This work combines an optimized Convolutional Long Short-Term Memory (ConvLSTM) model for capturing temporal and spatial patterns in physiological data with an adaptive Fuzzy Logic-based threshold adjustment to address these challenges. The results obtained with an accuracy of 99.38%, a recall of 99.50%, and an AUC of 0.996 show an acceptable reduction in false- positive rates. The proposed work dominantly reduces the false positives with improved effectiveness for real-time wearable WBAN.