Scholarlink

<p>Wireless Sensor Networks (WSNs) allow real-time patient monitoring by wirelessly transmitting health and environmental data, thereby improving healthcare management. However, accurate disease prediction remains challenging due to the limits of conventional monitoring methods and the heterogeneous nature of medical data. This research hypothesizes that integrating optimized deep learning with WSNs can significantly enhance patient monitoring accuracy and reliability. To address this, a Deep Learning-enabled Wireless Sensor Network for monitoring abnormal activities and patient health in healthcare environments (WSN-MA-PHE-TDGNN) is proposed. Input data from the Stroke Prediction Dataset is pre-processed using a Bilinear Double-Order Filter (BDOF) to remove noise and handle missing values. Adaptive Support Vector-Borderline Synthetic Minority Over-sampling Technique (ASV-SMOTE) balances the dataset by generating synthetic minority samples, followed by feature extraction via Refined Linear Chirplet Transform (RLCT). The Temporal Dynamic Graph Neural Network (TDGNN) classifies patient health as normal or stroke, with weight optimization performed by Carpet Weaver Optimization (CWO). The proposed WSN-MA-PHE-TDGNN method is implemented and its efficacy is evaluated in Python, utilizing multiple performance metrics. Experimental results show that WSN-MA-PHE-TDGNN achieves 99.21% Accuracy, 98.93% Precision, and 98.19% Recall, and 99.18% F1-Score, outperforming existing methods.</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

A Novel Deep Learning-Enabled Wireless Sensor Network for Monitoring Abnormal Activities and Patient Health in Healthcare Environments

  • 期刊论文

  • 发表日期：

    2026年3月17日

查看全文
  • Shweta S. Kaddi,
  • K S Rajeshwari,
  • H Pooja

摘要

Wireless Sensor Networks (WSNs) allow real-time patient monitoring by wirelessly transmitting health and environmental data, thereby improving healthcare management. However, accurate disease prediction remains challenging due to the limits of conventional monitoring methods and the heterogeneous nature of medical data. This research hypothesizes that integrating optimized deep learning with WSNs can significantly enhance patient monitoring accuracy and reliability. To address this, a Deep Learning-enabled Wireless Sensor Network for monitoring abnormal activities and patient health in healthcare environments (WSN-MA-PHE-TDGNN) is proposed. Input data from the Stroke Prediction Dataset is pre-processed using a Bilinear Double-Order Filter (BDOF) to remove noise and handle missing values. Adaptive Support Vector-Borderline Synthetic Minority Over-sampling Technique (ASV-SMOTE) balances the dataset by generating synthetic minority samples, followed by feature extraction via Refined Linear Chirplet Transform (RLCT). The Temporal Dynamic Graph Neural Network (TDGNN) classifies patient health as normal or stroke, with weight optimization performed by Carpet Weaver Optimization (CWO). The proposed WSN-MA-PHE-TDGNN method is implemented and its efficacy is evaluated in Python, utilizing multiple performance metrics. Experimental results show that WSN-MA-PHE-TDGNN achieves 99.21% Accuracy, 98.93% Precision, and 98.19% Recall, and 99.18% F1-Score, outperforming existing methods.