基于心电-脉搏波的心血管疾病识别研究
|
摘要
为实现心血管疾病的早期筛查,降低心血管疾病临床检测的成本。本研究基于上肢脉搏波传导速度(PWV)及脉搏波相关血液动力学基础理论,采集了总计51人的脉搏波与心电信号数据,提取了包括3种PWV和脉搏波特征参数总计16个特征参数,将不同的PWV与脉搏波特征组成3个样本特征数据集,分别建立了基于K近邻学习(KNN)和支持向量机(SVM)的心血管疾病识别模型。KNN模型分类准确率为66.28%,SVM模型分类准确率为84.3%,并通过对比不同PWV对模型性能的影响,确定了用于血管评估的最优脉搏波传导速度pwvm。研究表明基于SVM建立的分类模型对心血管疾病识别有一定可靠性,为低成本的心血管疾病早期筛查提供了新思路,也为穿戴式心血管系统监测提供了基础。
To achieve an early screening of cardiovascular diseases and reduce the cost of clinical detection of cardiovascular disease, a research based on pulse wave velocity(PWV) in the upper extremities and the hemodynamic theory related to pulse waves is performed. The pulse waves and electrocardiogram data of 51 volunteers were collected, and 16 feature parameters including 3 types of PWV and pulse wave features were extracted. Three sample feature data sets which are composed of different PWV and pulse wave features are used to establish two different cardiovascular diseases recognition models based on K-nearest neighbor(KNN) or support vector machine(SVM). The clssification accuracy of KNN and SVM models is 66.28% and 84.3%,respectively. By comparing the effects of different PWV on the performance of models, the optimal pwvm for vascular assessment is determined. The research results show that the SVM model is reliable in the cardiovascular disease recognition, providing a new idea for the low-cost and early screening of cardiovascular diseases and providing a basis for wearable cardiovascular system monitoring.
To achieve an early screening of cardiovascular diseases and reduce the cost of clinical detection of cardiovascular disease, a research based on pulse wave velocity(PWV) in the upper extremities and the hemodynamic theory related to pulse waves is performed. The pulse waves and electrocardiogram data of 51 volunteers were collected, and 16 feature parameters including 3 types of PWV and pulse wave features were extracted. Three sample feature data sets which are composed of different PWV and pulse wave features are used to establish two different cardiovascular diseases recognition models based on K-nearest neighbor(KNN) or support vector machine(SVM). The clssification accuracy of KNN and SVM models is 66.28% and 84.3%,respectively. By comparing the effects of different PWV on the performance of models, the optimal pwvm for vascular assessment is determined. The research results show that the SVM model is reliable in the cardiovascular disease recognition, providing a new idea for the low-cost and early screening of cardiovascular diseases and providing a basis for wearable cardiovascular system monitoring.
引文
[1]王文,朱曼璐,王拥军,等.中国心血管病报告2012概要[J].中国循环杂志, 2013, 28(6):408-412.WANG W, ZHU M L, WANG Y J, et al. The outline of Chinacardiovascular disease report 2012[J]. Chinese Circulation Journal,2013, 28(6):408-412.
[2]陈伟伟,王文,隋辉,等.《中国心血管病报告2016》要点解读[J].中华高血压杂志, 2017, 25(7):605-608.CHEN W W, WANG W, SUI H, et al. The key points of Chinesecardiovascular disease report 2016[J]. Chinese Journal ofHypertension, 2017, 25(7):605-608.
[3]隗玮,张存泰,吕美霞,等.无创臂踝脉搏波速度的特征及与心血管危险因素的关系[J].中华老年多器官疾病杂志, 2017, 16(4):274-278.WEI W, ZHANG C T, LüM X, et al. Features of noninvasive brachial-ankle pulse wave velocity and their relationship with cardiovascularrisk factors[J]. Chinese Journal of Multiple Organ Diseases in theElderly, 2017, 16(4):274-278.
[4] CAVALCANTE J L, LIMA J A, REDHEUIL A, et al. Aorticstiffness:current understanding and future directions[J]. J AmColl Cardiol, 2011, 57(14):1511-1522.
[5]朱妙章,唐朝枢,袁文俊,等.心血管生理学基础与临床[M].北京:高等教育出版社, 2011:198-199.ZHU M Z, TANG C S, YUAN W J, et al. The basis and clinic ofcardiovascular physiology[M]. Beijing:Higher Education Press, 2011:198-199.
[6]汪泉,王小飞,王蕾.脉搏波传导速度应用于心血管疾病研究进展[J].中国老年学杂志, 2014, 35(6):1720-1722.WANG Q, WANG X F, WANG L. Application of pulse wave velocityin cardiovascular diseases[J]. Chinese Journal of Gerontology, 2014,35(6):1720-1722.
[7] ALTY S R, ANGARITA-JAIMES N, MILLASSEAU S C, et al.Predicting arterial stiffness from the digital volume pulse waveform[J]. IEEE Trans Biomed Eng, 2007, 54(12):2268-2275.
[8] WU H T, LIN B Y, YANG C C, et al. Assessment of vascular healthwith photoplethysmographic waveforms from the fingertip[J]. IEEEJ Biomed Health Inform, 2017, 21(2):382-386.
[9]张丽娜,周润景,武佩,等.基于心电、脉搏波信号的动脉硬化无创检测[J].生物医学工程学杂志, 2016, 33(4):631-638.ZHANG L N, ZHOU R J, WU P, et al. Study on non-invasive detectionof atherosclerosis based on electrocardiogram and pulse wave signals[J]. Journal of Biomedical Engineering, 2016, 33(4):631-638.
[10]王开隽,张梅梅,叶继伦,等.基于指-趾脉搏波传导速度的无创动脉硬化检测系统设计[J].中国医学物理学杂志, 2016, 33(7):733-738.WANG K J, ZHANG M M, YE J L, et al. Design of non-invasiveatherosclerosis measurement system based on finger-toe pulse wavevelocity[J]. Chinese Journal of Medical Physics, 2016, 33(7):733-738.
[11]刘娜,李苗,赵静,等.单点脉搏波传导速度与冠状动脉狭窄数量的相关性研究[J].西安交通大学学报(医学版), 2015, 36(6):802-805.LIU N, LI M, ZHAO J, et al. Correlation between single point pulsewave velocity and coronary artery stenosis quantity[J]. Journal ofXi'an Jiaotong University(Medical Sciences), 2015, 36(6):802-805.
[12]梁宇彤,杨勇,陈鹏飞,等.动脉整体顺应性与颈-股脉搏波传播速度的相关性[J].中华医学超声杂志, 2017, 14(9):671-675.LIANG Y T, YANG Y, CHEN P F, et al. Correlation of arterial totalcompliance and carotid-femoral pulse wave velocity:anechocardiographic study[J]. Chinese Journal of Medical Ultrasound,2017, 14(9):671-675.
[13]ALLEN J. Photoplethysmography and its application in clinicalphysiological measurement[J]. Physiol Meas, 2007, 28(3):R1.
[14]AHN J M. New aging index using signal features of bothphotoplethysmograms and acceleration plethysmograms[J].Healthc Inform Res, 2017, 23(1):53-59.
[15]ELGENDI M. On the analysis of fingertip photoplethysmogram signals[J]. Curr Cardiol Rev, 2012, 8(1):14-25.
[16]MOHANALAKSHMI S, SIVASUBRAMANIAN A, SWARNA-LATHA A. Predicting arterial stiffness from physiologicalcharacteristics of photoplethysmography signals quantifiedthrough second derivative[J]. Indian J Sci Tech, 2017, 10(12):1-8.
[17]周谭琪,梁永波,刘桂勇,等.基于Relief算法的心血管疾病辅助诊断研究[J].生物医学工程学杂志, 2017, 34(4):535-542.ZHOU T Q, LIANG Y B, LIU G Y, et al. Study on aided diagnosis forcardiovascular diseases based on Relief algorithm[J]. Journal ofBiomedical Engineering, 2017, 34(4):535-542.
[18]周志华.机器学习[M].北京:清华大学出版社, 2016:225-226.ZHOU Z H. Machine learning[M]. Beijing:Tsinghua UniversityPress, 2016:225-226.
[2]陈伟伟,王文,隋辉,等.《中国心血管病报告2016》要点解读[J].中华高血压杂志, 2017, 25(7):605-608.CHEN W W, WANG W, SUI H, et al. The key points of Chinesecardiovascular disease report 2016[J]. Chinese Journal ofHypertension, 2017, 25(7):605-608.
[3]隗玮,张存泰,吕美霞,等.无创臂踝脉搏波速度的特征及与心血管危险因素的关系[J].中华老年多器官疾病杂志, 2017, 16(4):274-278.WEI W, ZHANG C T, LüM X, et al. Features of noninvasive brachial-ankle pulse wave velocity and their relationship with cardiovascularrisk factors[J]. Chinese Journal of Multiple Organ Diseases in theElderly, 2017, 16(4):274-278.
[4] CAVALCANTE J L, LIMA J A, REDHEUIL A, et al. Aorticstiffness:current understanding and future directions[J]. J AmColl Cardiol, 2011, 57(14):1511-1522.
[5]朱妙章,唐朝枢,袁文俊,等.心血管生理学基础与临床[M].北京:高等教育出版社, 2011:198-199.ZHU M Z, TANG C S, YUAN W J, et al. The basis and clinic ofcardiovascular physiology[M]. Beijing:Higher Education Press, 2011:198-199.
[6]汪泉,王小飞,王蕾.脉搏波传导速度应用于心血管疾病研究进展[J].中国老年学杂志, 2014, 35(6):1720-1722.WANG Q, WANG X F, WANG L. Application of pulse wave velocityin cardiovascular diseases[J]. Chinese Journal of Gerontology, 2014,35(6):1720-1722.
[7] ALTY S R, ANGARITA-JAIMES N, MILLASSEAU S C, et al.Predicting arterial stiffness from the digital volume pulse waveform[J]. IEEE Trans Biomed Eng, 2007, 54(12):2268-2275.
[8] WU H T, LIN B Y, YANG C C, et al. Assessment of vascular healthwith photoplethysmographic waveforms from the fingertip[J]. IEEEJ Biomed Health Inform, 2017, 21(2):382-386.
[9]张丽娜,周润景,武佩,等.基于心电、脉搏波信号的动脉硬化无创检测[J].生物医学工程学杂志, 2016, 33(4):631-638.ZHANG L N, ZHOU R J, WU P, et al. Study on non-invasive detectionof atherosclerosis based on electrocardiogram and pulse wave signals[J]. Journal of Biomedical Engineering, 2016, 33(4):631-638.
[10]王开隽,张梅梅,叶继伦,等.基于指-趾脉搏波传导速度的无创动脉硬化检测系统设计[J].中国医学物理学杂志, 2016, 33(7):733-738.WANG K J, ZHANG M M, YE J L, et al. Design of non-invasiveatherosclerosis measurement system based on finger-toe pulse wavevelocity[J]. Chinese Journal of Medical Physics, 2016, 33(7):733-738.
[11]刘娜,李苗,赵静,等.单点脉搏波传导速度与冠状动脉狭窄数量的相关性研究[J].西安交通大学学报(医学版), 2015, 36(6):802-805.LIU N, LI M, ZHAO J, et al. Correlation between single point pulsewave velocity and coronary artery stenosis quantity[J]. Journal ofXi'an Jiaotong University(Medical Sciences), 2015, 36(6):802-805.
[12]梁宇彤,杨勇,陈鹏飞,等.动脉整体顺应性与颈-股脉搏波传播速度的相关性[J].中华医学超声杂志, 2017, 14(9):671-675.LIANG Y T, YANG Y, CHEN P F, et al. Correlation of arterial totalcompliance and carotid-femoral pulse wave velocity:anechocardiographic study[J]. Chinese Journal of Medical Ultrasound,2017, 14(9):671-675.
[13]ALLEN J. Photoplethysmography and its application in clinicalphysiological measurement[J]. Physiol Meas, 2007, 28(3):R1.
[14]AHN J M. New aging index using signal features of bothphotoplethysmograms and acceleration plethysmograms[J].Healthc Inform Res, 2017, 23(1):53-59.
[15]ELGENDI M. On the analysis of fingertip photoplethysmogram signals[J]. Curr Cardiol Rev, 2012, 8(1):14-25.
[16]MOHANALAKSHMI S, SIVASUBRAMANIAN A, SWARNA-LATHA A. Predicting arterial stiffness from physiologicalcharacteristics of photoplethysmography signals quantifiedthrough second derivative[J]. Indian J Sci Tech, 2017, 10(12):1-8.
[17]周谭琪,梁永波,刘桂勇,等.基于Relief算法的心血管疾病辅助诊断研究[J].生物医学工程学杂志, 2017, 34(4):535-542.ZHOU T Q, LIANG Y B, LIU G Y, et al. Study on aided diagnosis forcardiovascular diseases based on Relief algorithm[J]. Journal ofBiomedical Engineering, 2017, 34(4):535-542.
[18]周志华.机器学习[M].北京:清华大学出版社, 2016:225-226.ZHOU Z H. Machine learning[M]. Beijing:Tsinghua UniversityPress, 2016:225-226.