PSYCHO-SOMATIC EFFECTS ON HEART RATE VARIABILITY DYNAMICS

Abstract

In contemporary times, the heart rate variability (HRV) analysis has become an established non-invasive tool to monitor autonomic control of heart in normal, pathological states and other clinical settings of a subject. It is proven fact that heart rate variability is higher in normal and healthy subjects, whereas reduced HRV has been associated with myocardial infarction (MI), coronary artery disease, ischemic heart disease and others. The high prognostic ability of HRV tool in predicting impending cardiac disease and investigating ability into autonomic nervous system regulation under different physiological and pharmacological interventions has enhanced its importance for further exploration. This can facilitate in taking timely and appropriate disease management and treatment measures. However, the HRV analysis and other related signal processing techniques and methodologies are still inaccurate, and there is a scope in their improvement for enhancing the accuracy of prognosis and interpretation. These are continuously evolving towards fulfilling quality healthcare needs. The efficient detection of R-peak is extremely important for the formation of reliable heart rate variability signal (RR-interval tachogram). The HRV indices are evaluated using the RR-tachogram. The spectral methods are reliable and popular for HRV analysis, but stationarity considerations of RR-interval time series while applying these methods for analysis introduce errors in parameter quantification. Still, the inherent advantages of spectral methods will always keep these methods highly relevant for future studies on heart rate variability. The autonomic nervous system regulates the heart beats through its sympathetic and para sympathetic nervous systems to maintain homeostasis. The heart rate is modulated by alterations in sympathetic and parasympathetic tones of autonomic control. The increase in sympathetic tone increases the heart rate, whereas increased parasympathetic or vagal tone decreases the heart rate. So, the synergistic action of these two branches (sympathetic and parasympathetic) characterizes heart rate variability (HRV), which refers to beat-to-beat alteration in heart rate. Moreover, the heart rate at any instant is resultant of many reflexes on the vagal and sympathetic centers such as baroreceptors, chemoreceptors, lung hyperinflation, atrial receptors and aortic chemoreceptors. There can be abrupt changes in heart rate in response to physical or mental stress and exercise. Even in the absence of external stimuli, the normal heart rate continuously varies during the day and night. XIX The RR-interval tachogram contains hidden variability information regarding autonomic control and its two controlling branches of sympathetic and parasympathetic activities. The RR-tachogram is re-sampled at appropriate frequency prior to its spectral analysis for the evaluation of HRV indices. The main spectral bands of interest in short term recording are referred to as the very-low frequency (VLF) band (< 0.04 Hz), the low frequency (LF) band (0.04-0.15 Hz) and the high frequency (HF) band (0.15-0.40 Hz). The VLF in the HRV spectra has been associated with thermoregulatory variations. The physiological interpretation of LF is controversial as both the sympathetic and parasympathetic contributions can be involved in this activity. The HF component corresponds to parasympathetic (vagal) activity. The ratio of the power contained in the LF and HF components has been used as a measure of the sympathovagal balance. The spontaneous fluctuations in the heart rate occur at very specific frequencies closely related to physiological functioning of the system, which is referred to as intrinsic oscillators. The intrinsic oscillators arise from respiratory sinus arrhythmia (RSA) with a frequency usually centered at 0.25 Hz in high frequency range (0.15Hz-0.40Hz), from baro-receptor loop (Mayer-waves) at 0.10 Hz in low frequency range (0.04Hz-0.15Hz), and from thermoregulatory modulations in very low frequency range (< 0.04 Hz). However, the location of these intrinsic oscillators may vary as per the subject's physiological status. After dealing with general introduction regarding heart rate variability and related physiology and technical literature, the first stage of the work presented here deals with the selection of appropriate tools and methodologies for heart rate variability analysis. The medical ethics guidelines were followed for recording the electrocardiogram (ECG) of human subjects. The ethical clearance was taken from the medical ethics committees of respective hospitals prior to recording of data. The technical requirements for recording and analysis i.e. duration of recording, data acquisition sampling frequency and ECG data of standard lead II was finalized for short term analysis of the HRV. The quality of heart rate variability spectral analysis depends on a series of accurate RRintervals forming RR-interval tachogram. This further depends upon accurate and efficient detection of R-peaks. This makes the QRS detection as the most important part of the heart rate variability analysis. The QRS detection algorithm based upon wavelet decomposition and adaptive thresholding was developed using existing wavelet functions of db6, db3, haar and bior2.2. These existing wavelet functions have been chosen for the detection algorithm as XX their scaling function (which describes shape of wavelet function) is closely similar to QRS complex. The algorithm was tested on varying data lengths of standard data bases (Fantasia data base of normal and healthy subjects and MIT/BIH data base of arrhythmia patients) and self recorded ECG signals of normal and healthy subjects and patients under disease stress. The R-peak detection efficiency of 99.59%, 99.48%>, 99.69%> and 99.61%) wasobserved using existing wavelet functions of db6, db3, haar and bior2.2 respectively. So, based upon the experience and wavelet selection guidelines, a new wavelet function has been proposed, which closely resembles QRS complex. It has given the best performance of 99.99% detection efficiency. This performance has been quoted after applying and testing the technique on large data sets of normal and patients with cardiac abnormalities as given in Fantasia data base, MIT/BIH arrhythmia data base and self recorded data. The detection of R-peaks affects HRV parameter evaluation and their interpretation. The Rpeaks were detected in standard annotated Fantasia data base for HRV evaluation. The algorithm using above new wavelet has detected correct R-peak locations and these have been verified in reference to available annotated RR-series of this record number. Comparison of HRV parameters using db6, db3, haar, bior2.2 and proposed newwavelet has been made. It has been shown that HRV parameters are affected by the correct beat detection of R-peaks. So, correct detection of R-peaks affects the HRV analysis and interpretation of results. HRV parameter variations with the aging of human male subjects has been made which may facilitate to establish normal limits of HRV parameters in three different age groups of a life span (18-30 years, 30-45 years and 45-60years). The ECG data of sixty seven normal and healthy male subjects in these age groups was recorded at the Institute Hospital of Indian Institute of Technology Roorkee after prior ethical clearance by the Hospital Ethics Committee. The written consent was taken from each subject and then they were medically examined by the expert physicians. ECG lead II data was recorded at 500 Hz sampling frequency alongwith medical history, if any. Five minutes of clean ECG data was used for HRV analysis after manual inspection, such that the considered data does not contain ectopy beats and is free of noise. The proposed new wavelet based QRS detection algorithm was used to form RR-tachogram for HRV spectral analysis which provides highly accurate HRV indices. The HRV has been evaluated using both the non-parametric (FFT) and parametric (AR) methods of analysis. It has been observed that the trend of corresponding HRV indices XXI with these two spectral methods remains same, but their absolute values may differ. The trend of variations in HRV indices was observed as measure based, but most of the HRV indices decline with age. However, their normalized contribution towards autonomic control activity has clearly shown the HRV indices changes with aging process. The normalized LF increases with increasing age, whereas normalized HF declines with aging which is also supported by decrease in RMSSD. This shows the dominance of sympathetic activity in autonomic control with increasing age. This trend is also indicated by increasing LF/HF ratio with aging. The heart rate (HR) is highest during active stage of psychosomatic activities in life, which can be roughly varying from 30-45 years of age, as assumed in our present study. The decrease in SDNN with increasing age shows the decline in total variability. It has been observed that the HRV indices computed with autoregressive (AR) modeling with model order determined by final prediction error (FPE) method differ in comparison to those evaluated with FFT technique. The heart rate variability indices evaluated with FFT approach have given the best results for 256 seconds of data using Hann window. However, this method suffers from spectral leakage effects due to windowing and this spectral leakage masks the weak signals present in the data. The autoregressive method avoids this spectral leakage and provides better frequency resolution than non-parametric (FFT) methods. The frequency resolution in AR spectral response is mostly affected by model order than the window length of the data. Moreover, it also offers the advantages of smooth power spectral density curve, easy identification of central frequencies in spectral bands corresponding to autonomic generators which arise due to thermoregulatory variations, baroreflex and respiratory sinus arrhythmia that is why many cardiologists prefer AR model to see concentrations of sympathetic and parasympathetic components. Therefore, a study has been made which pertains to the optimization of AR model order. In this, if the highest order of model is evaluated for total prediction error (TPE) of +1% more than error computed with Akaike's final prediction error (FPE) criteria for model order evaluation, then this order of AR model provides heart rate variability indices which are closely matched to heart rate variability indices evaluated with FFT method, without sacrificing the inherent advantages of AR modeling. However if model order is increased more than this optimized order, then HRV indices again starts deviating from those calculated with FFT. It has also been verified that the HRV indices remains nearly same with variable length of ECG signal. So, this optimized autoregressive model can be used for spectral analysis of the heart rate variability even for shorter length of ECG recording. This way, the problems pertaining to patient's XXll uncomfortability, which may lead to more undesired fluctuations due to limb and body movements, more number of ectopy beats and incorrect evaluation of heart rate variability indices. The applicability of this method on shorter data length may be useful in mobile settings. Psychosomatic effect is regarding constant and inseparable interaction of psyche (mind) and soma (body). Psychosomatic disease is thought to be made worse by mental factors, such as stress, anxiety, eczema, stomach ulcers and high blood pressure. So, another study has been made on the heart rate variability analysis of psychosomatic patients who were suspected to be suffering from some cardiac abnormality. This study was made under Memorandum of Understanding (MoU) signed between the Indian Institute of Technology Roorkee (India) and the Himalayan Institute and Hospital Trust, Dehradun (India) for promoting research and development in frontal areas of health care with an expert cardiologist agreeing on joint supervision of this work. The ethical clearance for this research work was provided by Hospital Ethics Committee of HIHT, Dehradun after thoroughly examining the research problem and enquiring about some relevant issues. The patients attending the outpatient department at the Cardiology Department, HIHT, Dehradun were clinically examined by expert cardiologists and after taking written consent of patients, the ECG data of total fortysix patients was recorded at 500 Hz sampling frequency using BIOPAC MP 100 system. Five minutes of clean, free from any ectopy beats and noise ECG data was used for HRV analysis after manual inspection, although data was recorded for a total of fifteen minutes. The Rpeaks were detected using proposed new wavelet based algorithm thereby forming highly accurate RR-tachogram. The HRV indices were evaluated using proposed optimized autoregressive modeling and conventional time domain indices of SDNN, SDSD, RMSSD and HR. As per the opinion of expert cardiologists, the patients were classified in four groups (Class SI, Class S2, Class S3 and Class S4) on the basis of the severity of cardiac abnormality based upon symptoms. In this respect, an HRV indices based artificial neural network (ANN) cardiac health classifier has been proposed with one hidden layer and an output layer. The HRV indices data of twenty eight patients was used for training of this proposed ANN based cardiac health classifier, whereas the HRV indices data of remainig eighteen patients was used for testing of this classifier. So, this perceptron model is simple and has ability to classify the patients' cardiac health based upon his HRV indices. This study is useful to classify the patients based on HRV analysis and physical symptoms of the patient. XXlll The modern technology and boom in information technology has prompted the researchers to step up their activities to harness the ancient alternate medication system of yoga therapy. The recent studies on yoga therapy have demonstrated its therapeutic potential to influence the body's physical and mental health. So, another study has been undertaken in this work, in which heart rate variability dynamics has been evaluated during four yogic-based mental states which includes state of random thoughts (cancalata), focusing (ekagrata), meditation with focusing (dharana) and meditation (dhyana). The object of meditation used in this study is a holy symbol of'Om'. The nomenclature of cancalata, ekagrata, dharana and dhyana has been acquired from Vedic literature. The electrocardiogram data of thirty four experimental volunteers was acquired at the Swami Vivekanand Yoga Research Foundation, Bangalore, India and was shared with prior permission and sharing of common interests. The selected volunteers in the age range of 22 to 38 years (mean ages32 years, mean height s 164 cms) were regular yoga practitioners having 3-12 years of meditation experience and were used to practice 3-5 times/week. The signal quality in four recorded signal was not adequate due to artifacts and undesired subject's movements due to uncomfortability, so in total, good quality ECG data of thirty subjects were used for analysis. The ECG data of more than 32 minutes duration was recorded, which includes 5 minutes of pre-session, and four sessions each of five minutes duration during given yogic-based mental states and five minutes of post-session recording. Under this, three studies have been made to investigate heart rate variability dynamics under varying neuro-humoral interactions. The heart rate variability indices have been evaluated using proposed optimized AR modeling. It is hypothesized that state of meditation (dhyana) is expected to lead human body towards homeostasis, if subject is under state of random thoughts (cancalata). So, to study comparison of state of meditation (dhyana) and random thoughts (cancalata) effects on autonomic control, the heart rate variability indices were evaluated. The intervention in case of dhyana is imagination of a holy symbol 'Om', followed by Om chanting and finally merging into it. The intervention in case of cancalata was listening of illogical FM radio play. It has been observed that the sympathetic tone (LF power) becomes higher under cancalata state of mind. The autonomic balance (LF/HF ratio) is dominated by sympathetic tone, which leads to reduced heart rate variability under cancalata state of mind, whereas the heart rate variability is higher under 'Om' meditation in comparison to cancalata state of mind. The thermo regulatory variations are increased under Cancalata, whereas these variations are reduced under dhyana. The mean hear rate is reduced under 'Om' mediation, XXIV whereas it becomes higher under random thoughts. So, the neuro-humoral body visceral regulation has been observed under control in the state of dhyana. The second study is regarding comparison of the heart rate variability under the state of concentration (ekagrata) (i.e. when subjects are intervened by multidirectional and logically connected thoughts on a single subject) and meditation with focusing (dharana) i.e. when there is unidirectional thought on a single subject). The intervention in case of ekagrata state used was play of logical lecture on cyclic meditation, whereas in case of dharana intervention was to imagine the holy symbol 'Om'. The heart rate variability is higher under ekagrata state of mind than dharana. The thermo-regulatory variations are higher under ekagrata sessions, which may be attributed to increased visceral body dynamics, but in post ekagrata session thermoregulatory variations subside. The thermoregulatory variations under dharana sessions become prominent, but are inflated in post dharana session. The mean heart rate is reduced more during dharana sessions, but this decrease in heart rate is more in post-ekagrata session. The third study is regarding the variation of comparison of HRV indices, when the subject passes through different states of mind from cancalata, ekagrata, dharana to dhyana in pursuit of state of samadhi. The sympathetic activation during cancalata state becomes high and vagal (parasympathetic) control is weakened. The Ekagrata state has observed to strengthen the vagal tone of autonomic control. The dharana state may induce unmatched cardiovascular response which leads to poor heart rate variability. It persists even in its post session. The Om meditation effect during dhyana session has revealed marked rise in heart rate variability, which can be gauzed by strengthening of vagal tone. The total variability is also increased as observed in total power. Moreover, the heart rate is lowest under meditative state, which can be attributed to paradoxically induce quiescent heart rate dynamics. So, in totality, the autonomic control is differentially affected in a particular mental state, which may be beneficial in conditions, where particular variability is blunted due to some disease. These studies will be useful in appreciating the therapeutic affects of these various mental states and benefits of meditation (dhyana) on cardiac health care.