一种基于心率和深层学习的心电图分类算法

目的研究适用于远程医疗服务系统、体检中心和临床应用的心电图(electrocardiogram,ECG)正异常分类算法。方法首先,通过心率筛除异常数据。然后,对于心率判为正常的心电图,采用LCNN对心电图再次进行正异常分类,并对多个LCNN的分类结果进行融合。结果在15万多条记录的临床数据集上测试,取得了84.77%的准确率,85.19%的灵敏度和84.45%的特异性。结论该实验结果优于对照文献,同时对应用于远程医疗和体检中心的计算辅助分析方法具有一定的参考价值。

An ECG Classification Algorithm Based on Heart Rate and Deep Learning

Objective To study the electrocardiogram (ECG) normality vs.abnormality classification algorithm applicable to remote medical service system,physical examination center and clinical application.Methods First,the abnormal data was eliminated through the calculation of heart rate and the ECG with normal heart rate was selected.For the ECG with normal heart rate,normality vs.abnormality classification was carried out with lead convolution neural network (LCNN),and then the classification results were fused.Results The test of over 150,000 clinical records showed that the accuracy was 84.77％,the sensitivity was 85.19％ and the specificity was 84.45％.Conclusion The result is better than the literature.It can serve as a reference for the computer-aided analysis method in remote medical and physical examination center.