Multi-Modal Biomedical Signal Analysis for COPD Early Warning: A Wearable and Wireless Communication-integrated Solution

作者：Tong You, Tao Dong, Zhaochu Yang, Wangang Zhu, Yao Lu, Chenghui Yu, Bingde Zou, 甘小彪, Zhi Xu, Xue Han · 年份：2026 · DOI：10.1109/cisce69494.2026.11504869 · 研究领域：Non-Invasive Vital Sign Monitoring、ECG Monitoring and Analysis、Phonocardiography and Auscultation Techniques

Chronic Obstructive Pulmonary Disease (COPD) requires continuous monitoring to manage exacerbations and prevent complications. This study presents an innovative home-based monitoring system that integrates multi-sensor fusion, edge computing, and artificial intelligence to enable intelligent, real-time assessment of COPD status. The system employs flexible strain sensors, a miniature microphone, and an accelerometer to non-invasively collect multimodal physiological data, including chest-abdominal movement, respiratory sounds, and activity levels. Data are transmitted via Bluetooth Low Energy (BLE 5.0) to edge nodes (e.g., smartphones) for noise reduction and feature extraction, significantly reducing data volume and power consumption. A machine learning model (XGBoost) is then utilized to classify patients into three clinical phases—stable, subacute exacerbation, and acute exacerbation—and to provide early warnings for risks such as hypoxia. The system supports dual-end applications (patient and physician) for visualization and remote management, forming a closed-loop monitoring framework from data acquisition to clinical decision support. This work demonstrates a feasible, efficient, and patient-centered technical pathway for the remote management of COPD, with implications for scalable intelligent healthcare solutions.