穿戴式心电信号处理的几个关键问题研究及在房颤检测中的应用

随着大健康战略的实施和移动技术的发展，穿戴式生理信号的实时测量与处理成为现实.真正做到人体几个关键指标（如心电、脉搏、血压）在人体自然状态下的实时监控，可以早期发现50%以上人类健康问题。但实时和动态生理信号监控的数据量巨大、噪声强和大量的信息冗余，使得实时处理相对静态异常困难，成为穿戴式设备走向临床的瓶颈。鉴于此，项目以穿戴式心电在房颤疾病的早期发现、术后评价为突破口，针对信号处理的老问题（去冗余、去噪声），基于SQI指数发展心电信号质量评估理论与方法，筛选临床有意义的信号段；研究信号的稀疏性和信息融合理论实现稳健RR序列构建；发现RR序列不同时间尺度上的生物特征，为RR序列研究从静态走向动态、从统计有效到真正临床提供创新的理论和方法；最后针对临床常见的房颤疾病，探索AI诊断理论，建立临床房颤的预测与术后评价模型。

As the implementation the health strategy and the development of mobile technology, real time measurement and processing of wearable physiological signals are to become a reality.If a few key indicators (ECG\Pulse\blood pressure) are monitoring in the body's natural state, more than 50% of the early human health problems can be solved, but the dynamic monitoring of physiological signals lead to great amount of data and big noise and information redundancy, that make this work extremely difficult than that in static. It become the bottleneck of wearable devices come to clinical application. This research is on several key problems of wearable ECG signal processing and its application in atrial fibrillation detection. Aiming at the old problems of signal processing (de-redundancy and de-noise), the ecg signal quality evaluation theory and method are developed based on SQI theory.The sparse signal and information fusion theory are studied to realize robust RR sequence construction.The biological characteristics of RR sequence on different time scales were discovered, which provided innovative theories and methods for the study of RR sequence, to push this technology into clinical. Finally, Aiming at the common clinical atrial fibrillation diseases, the AI diagnosis theory was explored, and the prediction and postoperative evaluation model of clinical atrial fibrillation was established.