使用HRV多特征参数和机器学习的心衰诊断方法研究  被引量:1

Detection of Heart Failure Based on Multi-Features from HRV and Machine Learning

在线阅读下载全文

作  者:张敏 赵兴群[1] ZHANG Min;ZHAO Xingqun(College of Bioscience and Medical Engineering,Southeast University,Nanjing Jiangsu 210096,China)

机构地区:[1]东南大学生物科学与医学工程学院,江苏南京210096

出  处:《电子器件》2021年第4期1005-1010,共6页Chinese Journal of Electron Devices

摘  要:心力衰竭是一种常见的严重心脏病,其诊断主要根据症状、体征及辅助检查等大量临床检查综合判断,昂贵又费时。而被广泛应用于各种心血管疾病诊断中的心电图具有无创、简便、经济等优点。心率变异性是从心电图中计算得到的逐次心跳周期的变化情况,是心血管疾病病情诊断及预防中一个有价值的指标。本研究选取心率变异性信号的时域分析、频域分析、非线性分析下14种参数作为多特征,使用机器学习中决策树、KNN、贝叶斯、SVM、随机森林等分类器,实现对心衰信号的识别。结果表明,模式识别方法均能有效检测心衰信号,其中,SVM方法准确率最高,可达98.81%,为自动诊断充血性心衰提供了一种有效的工具。Heart failure is a common serious heart disease.The diagnosis is mainly based on the comprehensive judgment of a large number of clinical examinations such as symptoms,signs and auxiliary examinations,which is expensive and time-consuming.The electrocardiogram,which is widely used in the diagnosis of various cardiovascular diseases,has the advantages of non-invasive,convenient,economical,and so on.Heart rate variability is the change of the heartbeat cycle calculated from the electrocardiogram,and it is a valuable indicator in the diagnosis and prevention of cardiovascular disease.In this study,14 parameters of time domain analysis,frequency domain analysis,and nonlinear analysis of heart rate variability signals are selected as multi-features,and the classifiers such as decision trees,KNN,Bayes,SVM,and random forest in machine learning are used to realize the recognition of heart failure signals.The results show that the pattern recognition methods can effectively detect heart failure signals.Among them,the SVM method has the highest accuracy rate of 98.81%,which provides an effective tool for automatic diagnosis of congestive heart failure.

关 键 词:心率变异性 多参数 心衰诊断 机器学习 

分 类 号:TN911.71[电子电信—通信与信息系统] R540.41[电子电信—信息与通信工程]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象