Heart Rate Variability Analysis Of Chronic Heart Failure Patients Based On Holter Data

Background:It was revealed that congestive heart failure (CHF) patients are often characterized by impaired autonomic nervous system (ANS), accompanied by the loss of sympathinhibition. As a noninvasive measure, heart rate variability (HRV) is used to assess the balance between sympathetic and vagal of ANS on the heart, which has been widely used for CHF. Based on the time interval series from ECG, the study is to investigate the behavior of ANS through the circadian rhythm of HRV analysis, which would provide a promising method to find application for noninvasive assessment of ANS state.Methods:One dataset from THEW as normal controls (n=189) and two datasets of CHF patients that are Congestive Heart Failure RR Interval Database and BIDMC Congestive Heart Failure Database from PhysioNet were selected in this study. According to NYHA class,44 CHF patients were divided into mild CHF group (NYHA Ⅰ-Ⅱ, n=12) and severe CHF group (NYHA III-IV, n=32). Two 2 h episodes of RR interval (RRI), representing day and night in resting state, were selected in each Holter record. Then, time domain analysis, AR model based frequency domain analysis, symbolic dynamic anslysis (SDA), detrended fluctuation analysis (DFA) and non-Gaussian analysis were calculated in each series to investigate the changes of HRV circadian rhythm and characteristics behavior of ANS for CHF patients.Meanwhile,5min periods were randomly selected from each 2h period. Then, the short-term time domain analysis, AR model based frequency domain analysis, symbolic dynamic anslysis (SDA), detrended fluctuation analysis (DFA) indexes were calculated. Through the statistical analyses, the indexes which reduce individual differences and highlight pathological factors were picked and used in Fisher linear discriminant analysis and support vector mechine analysis to discriminate normal people and CHF patients, as well as discriminate mild CHF patients and severe CHF patients.Result:The short-term time domain index (RRI), frequency domain index(LFn and HFn), SDA indexes (0V% and 2UV%) and DFA index (α1) existed significantly differences between day and night in normal group, but differences disappeared in the mild group and reoccurred in severe group. Moreover, circadian rhythm in severe group has a reverse trend compared with that in normal group, which relected circadian rhythm of sympathetic and vagal of CHF patients disappeared or reversed.The results demonstrated that there were significant differences for short-term fractal scaling exponent in the day (α1)d in any two groups among normal controls, mild CHF group and severe CHF group. Moreover, the declining trend of (α1)d (1.35±0.21,1.03±0.29 and 0.81±0.29, respectively) showed the change of heart rate dynamics from fractal properties towards random structure. In the meanwhile, significant differences existed for HFn in the day ((HFn)d) in any two groups among the abovementioned three groups. And the sustained increase of (HFn)d (23.89%±12.78%,37.22%±11.24% and 56.30%±15.28%, respectively) suggested the loss of reciporcal function between sympathetic and vagal branches. Using RRIn, (HFn)d and (α1)d, the sensitivity and specificity for discriminating normal people and CHF patients reached 90.91% and 92.06% through Fisher linear discriminant analysis, while the sensitivity and specificity for discriminating mild CHF patients and severe CHF patients were 84.38% and 100% through Fisher linear discriminant analysis. But through support vector mechine analysis, the sensitivity and specificity for not only normal people and CHF patients but also mild CHF patients and severe CHF patients are 100%.Conclusions:Combining nonlinear analysis and traditional methods, our results provided the basis for clinical use of HRV in monitoring the progress of CHF and testing therapeutic effects. Sympathetic overactivity and sympathovagal accentuated antagonism interaction impared the balance of ANS, which lead to the changes of circadian rhythm pattern. Meanwhile, the indexes which reduce individual differences and highlight pathological factors were picked using short-term HRV analysis to provided the basis for clinical use of HRV in monitoring the progress of CHF and testing therapeutic effects.