Predicting sudden cardiac death by analyzing its heart rate variability signal using Hilbert-Huang transform

• Main Authors: Tsun-Wei Tang, 湯存偉
• Other Authors: Woei-Chyn Chu
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/4n85aj

碩士 === 國立陽明大學 === 生物醫學工程學系 === 105 === According to the investigation done by the Ministry of Health and Welfare, heart attack is a major cause of death of elders in Taiwan. The sudden cardiac death onset leads to short rescue time, thus causes extremely high mortality rate. If we can detect the signs of heart attack beforehand, it may offer time to take a prompt medication action. The aim of this study is to investigate the relationship between heart rate variability (HRV) and sudden cardiac death by means of electrocardiography (ECG) analysis. The heart rate variability analysis used in this study includes time domain and frequency domain analysis, and the analysis in the frequency domain is performed by the Hilbert-Huang transform and Fourier transform. This study analyzed the ECG data measured minutes before sudden cardiac death onset, and the result shows that there were statistically significant differences. For the five minutes HRV analysis, nine of eleven features extracted from the HRV showed statistically significant differences between healthy adults and the adults who are on the verge of suffering sudden cardiac death. Furthermore, we trained those features by using the k-nearest neighbor algorithm as a classification tool. The accuracy rate of this classification model used to distinguish between the healthy adults and cardiac patients is 94.7%. And the accuracy rates analyzed using the ECG data measured more than five minutes, including ten minutes, thirty minutes and an hour, before onset is about 70-80%. This study reports our proposed method may detect sudden cardiac death minutes before its onset.