人口老龄化和慢性病给公共医疗健康系统带来了巨大压力，由于心血管疾病具有突发性和难以预警的特性，导致其死亡率一直居于首位。本项目基于心电体域网，研究心律失常信号的精准识别技术和实时可靠传输方法，突破心电信号时空多维特征难以即时分析，和异常信号易被干扰难以即时预警的技术瓶颈。项目研究内容主要包括：（1）研究一种将卷积神经网络和双向长短期记忆网络高度耦合的心律失常精准识别方法，联合学习心律的波形和时序两方面特征，提升心脏健康状态的识别精度；（2）针对心律失常预警数据、诊疗数据和常规数据，研究一种基于李雅普诺夫框架的分级传输优化方法，以确保心电信号在体域网传输过程中的时效性和可靠性。研究成果将形成基于体域网的慢性病监测支撑系统，可为患者提供即时、准确、可靠的心律失常监测及实时预警服务，缓解我国目前医疗资源分布不均，以及心血管疾病死亡率逐年升高的态势。

The aging population and chronic diseases have brought great pressure to the public health system. Due to that, cardiovascular disease is usually sudden and difficult to early warning, its mortality has always been the highest. Based on ECG-BAN, this research studies the accurate arrhythmia recognition and the real-time reliable transmission method, and tries to break through the technical bottleneck of analyzing the multidimensional space-time characteristics of ECG signal, solve the signal interference problem, and provide real-time early-warning services. The research mainly includes: (1) to study an accurate recognition method of arrhythmia, which is highly coupled with convolutional neural network and bidirectional long short-term memory network, and jointly learn the waveform and time sequence characteristics of arrhythmia, to improve the recognition accuracy of heart health state; (2) to study a method based on the Lyapunov frame for the arrhythmia early warning data, diagnosis data, and routine data to ensure the timeliness and reliability of ECG data transmission in BAN. The research results will form a chronic disease monitoring support system based on BAN, which can provide real-time, accurate, and reliable arrhythmia monitoring and early warning services for patients, and alleviate the uneven distribution of medical resources and the increasing trend of cardiovascular disease mortality in China.