一种利用心率检测心律失常的方法

An Arrhythmia Detection Method Based on Heart Rate Analysis

目的探索利用心率序列检测心律失常的方法。方法由MIT-BIH心律失常数据库的注释文件得到R-R间期序列及心拍标注信息。将心拍标注中的窦性心拍、束支传导阻滞、各类融合心拍、起搏心拍、未分类心拍定义为"正常心拍";将早搏及逸搏定义为"非正常心拍"。对原始R-R间期序列依次进行中值滤波与均值滤波,得到平滑R-R间期序列;原始R-R间期与平滑R-R间期的差值过大者为"离群"R-R间期,相应的心拍归为"非正常心拍"。结果 MIT-BIH心律失常数据库全部48例记录中,有35例记录的窦性心拍数量占总心拍数量的一半以上。这35例记录"非正常心拍"检测的敏感度为88.2%,阳预测率为91.3%;对其中不含房颤或房扑的28例记录"非正常心拍"检测的敏感度为94.2%,阳预测率为96.4%。结论利用心率序列可以识别多数心室期前收缩与心房期前收缩,从而降低人工判读心律失常的工作量。

Objective To explore the way of detecting arrhythmia by using the feature of heart rate. Methods R- R intervals and heartbeat type were obtained from the annotation of MIT-BIH Arrhythmia Database. Several heartbeat types from the database＇ s annotation were defined as ‘normal heartbeats＇ , including sinus beat, bundle branch block beat, fusion beat, paced beat, and unclassifiable beat. Heartbeat types of premature beat and escape beat were defined as ‘abnormal heartbeats＇. Median-filtering and average-filtering were applied to the original R-R intervals and then smoothed R-R intervals were obtained. There was significant difference be- tween some original R-R intervals and their corresponding smoothed R-R intervals, and the corresponding heartbeats were attributed as abnormal heartbeats. Results There were 35 out of 48 records of which the per- cent of sinus beat was more than 50%. A sensitivity of 88.2% and a positive prediction of 91.3 % were ob- tained on these 35 records for abnormal heartbeat detection. In addition, a sensitivity of 94.2% and a positive prediction of 96.4% were achieved on the 28 records without atrial fibrillation or atrial flutter. Conclusion The arrhythmias of atrial and ventricular premature beats can be identified by heart rate analysis thus lighten the burden of manual arrhythmia scanning.