Abstract
Acute myocardial infarction (AMI) onset when an interruption in the supply of myocardial oxygen and nutrients occurs and causes damage to the myocardium. The infarct area may hinder the normal contraction of the myocardium, increasing the loading of healthy parts of the heart, and even cause death of more cardiac cell. Finally, it may evolve into heart failure or other more severe conditions. Therefore, we want to find some parameters that can be used as a predictor of cardiac deterioration. In the thesis, according to the blocking blood vessel, AMI patients are divided into right coronary artery (RCA) group and left anterior descending (LAD) group. Every patient recorded 24 hours ECG during the acute stage (within 72 hours after AMI), 3 months post-MI (90 ± 14 days after AMI), 6 months (180 ± 28 days after AMI), and chronic phase (1 year). At each stage, we only extracted four hours data with waking period in heart rate variability (HRV) analysis, and then used Mann-Whitney U test to assess the difference in HRV parameters between AMI patients and the controls. The HRV results shown that there is a difference between AMI patients and the control subjects. At the chronic phase, the parameters of frequency domain and multiscale entropy (MSE) curve display the difference between AMI patients and the controls. At the acute stage, the long-term fractal scaling exponent (α2) in detrended fluctuation analysis (DFA) of AMI patients is significantly lower than that of the controls. The results also proved that the nonlinear methods of HRV such as MSE and DFA can provide some information that traditional parameters not showed. Since the HRV parameters, which had a significant difference between RCA group and the controls, are slightly different to LAD group, it confirms that the effect on the heart caused by the blocking of dissimilar blood vessels is not the same. Therefore, RCA group and LAD group should not be mixed into one group. Moreover, we want to find out some available parameters from acute stage ECG signal to predict the probability of AMI recurrence, heart failure or death. However, none of the patients died or developed heart failure during 1 year of follow-up. Therefore, we divided patients into two groups according to the left ventricular ejection fraction (LVEF) after AMI attack one year. The group with lower LVEF was regarded as a group of AMI patients who have a good status of the heart, and the other group was regarded as a group of AMI patients who have a better status of the heart. The HRV parameters, the standard deviation of all normal to normal intervals (sdNN), low frequency (LF), the slope of MSE curve in short scale (slope1-5) and DFA α2 were significantly different at the acute stage between two groups. Creatinine kinase -MB (CKMB), which is obtained from blood tests and related with the damaged area of the heart, also had a significant difference between two groups. In order to compare the discrimination ability of those parameters and CKMB, we used receiver operating characteristic (ROC) curve analysis. LF has the best discriminatory power (AUC=0.8051), the second one is slope1-5 (AUC=0.7721), and the third one is CKMB (AUC=0.739). Then we used LF, slope1-5 and CKMB to establish a logistic regression model, and the AUC of this model is 0.8235. It indicated that combining CKMB and HRV parameters can really enhance the discrimination rate.