论文部分内容阅读
本文研究了Relief特征选择方法在光电容积脉搏波(PPG)中的应用,分析寻找区分心血管疾病的指标,提出了一种辅助心血管疾病诊断的方法。通过收集40位志愿者的生理病理信息,并实时采集血压与指尖PPG波形数据,形成样本数据集。基于PPG波形,定义并提取了52个特征参数,通过特征选择Relief算法筛选出10个核心特征参数,形成最优特征子集,并分析它们对心血管疾病的影响。最后使用分类算法建模,对心血管疾病做出了辅助诊断,k邻近算法(k NN)模型对心血管疾病的预测正确率达到66.67%,支持向量机(SVM)模型对心血管疾病的预测正确率达到83.33%。结果表明:(1)年龄对心血管疾病辅助诊断最为重要;(2)最优特征子集元素特征为心血管健康状况评价与预测提供了重要依据。本研究表明,经Relief算法选择得到的最优特征子集为心血管疾病辅助诊断提供了更高的准确性。
In this paper, the application of Relief feature selection method in photoplethysmography (PPG) is studied. The aim of this study is to find out the index of distinguishing cardiovascular diseases and to propose a method to diagnose cardiovascular diseases. By collecting the physiological and pathological information of 40 volunteers and collecting blood pressure and fingertip PPG waveform data in real time, a sample dataset was formed. Based on PPG waveforms, 52 feature parameters were defined and extracted, and 10 core feature parameters were screened by feature selection Relief algorithm to form the optimal feature subset and analyze their impact on cardiovascular diseases. Finally, the classification algorithm was used to make a diagnosis of cardiovascular diseases. The prediction accuracy of the k-nearest neighbor algorithm (k NN) model for cardiovascular disease was 66.67%. The prediction of cardiovascular diseases by support vector machine (SVM) The correct rate reached 83.33%. The results showed that: (1) Age is most important for the diagnosis of cardiovascular diseases; (2) The characteristics of the optimal subset of features provide important evidence for the evaluation and prediction of cardiovascular health status. This study shows that the optimal subset of features selected by Relief algorithm provides a higher degree of accuracy for the diagnosis of cardiovascular disease.