Journal of Industrial and Intelligent Information Vol. 1, No. 2, June … · 2013-07-03 · cyclic meditation (CM) and supine rest (SR), showed decrease in LF power and LF/HF ratio

Effect of Ekagrata Meditation on Sympatho-Vagal

Balance-An Auto-Regressive Model Approach

Ramesh K. Sunkaria Department of Electronics & Comm. Engg., Dr. B R Ambedkar National Institute of Technology,

Jalandhar (Punjab), India

Email: [email protected]

Abstract—The ekagrata meditation state pertains to

channelizing the thoughts in one direction. In the present

work the effect of meditation state of ekagrata on heart rate

variability dynamics have been analyzed. For this, the

electrocardiogram was recorded in thirty (n=30)

experienced male yoga practitioners (mean age 32 years,

range 22-38 years, mean height 164 cms) in pre, during

and post-sessions of ekagrata meditation. The heart rate

variability was evaluated using optimized autoregressive

technique. The sympatho-vagal balance (LF/HF ratio)

became sympathetic tone (LF power) dominant, while vagal

tone (HF power) was observed to be significantly low

(p<0.01) under meditation state of ekagrata. The total

variability (Ptotal (ms2)) is significantly high (p<0.05) and

mean heart rate (HR (bpm)) is lowered under this

meditation state.

Index Terms—heart rate variability, ekagrata meditation,

optimized AR model, LF power, HF power.

I. INTRODUCTION

The previous studies on yoga therapy, breathing

exercises and meditation have demonstrated their

therapeutic potential to influence the autonomic nervous

system dynamics. Howorka K. et al. (1995) demonstrated

an immediate decrease of sympathetic and increase of

parasympathetic activity after yoga in a preliminary study

[1]. The heart rate variability study by Raghuraj P. et al.

(1998) in two selected breathing techniques, viz.

Kapalabhati (breathing at high frequency) and nadisudhi

(alternate nostril breathing) exhibited significant increase

in low frequency (LF) power and LF/HF ratio while high

frequency (HF) power was significantly lowered

following kapalabhati, but there were no significant

changes following nadisuddhi [2]. Peng C. K. et al. (1999)

reported extremely prominent heart rate oscillations

during Chinese Chi and Kundalini yoga meditation

techniques with slow breathing, which challenged the

very notion of meditation as only an autonomically

quiescent state [3]. Peng C. K. et al. (2003) in their later

study on heart rate variability in three forms of meditation

observed increased coherence between heart rate and

breathing during relaxation response and segmented

breathing when compared to baseline. The third form of

meditation i.e. breath of fire was marked by increase in

Manuscript received April 6, 2013; revised May 8, 2013.

mean heart rate with respect to baseline and a significant

decrease in coherence between heart rate and breathing

[4]. Vladimir S. and Ofer B. (2005) emphasized the

sympathetic nervous system dynamics to counter mental

stress and other external or internal disturbances effecting

visceral homeostasis [5]. Jovanov E. (2005) during very

slow yogic breathing found increased VLF frequencies,

respiratory sinus arrhythmia, LF/HF ratio and decreased

breathing frequency after exercise in comparison to state

before exercise [6]. Patil S. and Telles S. (2006) in their

study on two yogic based relaxation effects on HRV in

cyclic meditation (CM) and supine rest (SR), showed

decrease in LF power and LF/HF ratio during and after

CM, whereas HF power decreased. The heart rate

increased during CM but decreased after CM. However,

there was no change in SR state [7]. Travis F. et al. (2008)

tested the advantageous effect of transcendental

Meditation (TM) technique in overcoming stress in

college students [8]. Sunkaria et al. (2010) observed high

HRV in yogic practitioners in comparison to non-yogic

practitioners [9].

The present study evaluates heart rate variability

dynamics during meditation states of ekagrata. The effect

of this meditation state on autonomic control dynamics

may enhance the therapeutic utility in relevant cardiac

health conditions.

II. METHODS

A. Subjects

Thirty four experimental volunteers for

electrocardiogram (ECG) recording were selected from

Swami Vivekanand Yoga Research Foundation,

Bangalore, India. The selected volunteers were regular

yoga practitioners having 3-12 twelve years of meditation

experience and were used to practice 3-5 times/week.

They were in excellent health condition and were not

having even history of any cardiac or vascular disease.

This was also ascertained by a thorough medical

examination. The subjects were explained the

experimental procedures and that the recorded ECG

signal would be used for academic and research purpose

alone. An informed written consent was obtained from

each subject prior to recording. The lead-II ECG was

recorded in each of the thirty four male volunteers having

age in the range of 22-to -38 years (mean age 32 years,

102

Journal of Industrial and Intelligent Information Vol. 1, No. 2, June 2013

©2013 Engineering and Technology Publishingdoi: 10.12720/jiii.1.2.102-106

mean height 164 cms) under ekagrata state. The signal

quality in four recorded signals was contaminated by

artifacts especially due to subject’s movements, so in

total, good quality ECG data of thirty subjects were used

for analysis.

B. Inclusion/Exclusion Criteria

The normal and healthy yoga practitioners were

included in the present study after a thorough medical

examination and fulfilling certain physiological

requirements. Subjects having pulse rate > 100 bpm,

blood pressure > 160/100 and subjects suffering from

severe breathlessness were left out of the recording

process. The ECGs were checked and none of the

subjects had extra systoles.

C. Intervention of Ekagrata Meditation State and

Recording Protocols

Each of the selected yoga practitioners underwent ECG

recording protocol as per Table I under ekagrata state of

meditation with total duration of more than thirty minutes

comprising of pre, during (D1, D2, D3 and D4) and post

session with conditioning gaps. The interventional

meditation states of ekagrata are channelizing the

thoughts in one direction. All recording sessions of pre,

D1, D2, D3, D4 and post were of duration five minutes

each. The recording was made on subsequent days

between 05:00 and 08.30 hours in the morning or 05:00

and 08:30 hours in the evening to avoid outside

disturbance to cover all the subjects. The instructions to

all subjects were as per given in the Table I.

TABLE I. THE ECG DATA RECORDING PROTOCOLS

Sessions ekagrata

Pre (5 min) Sit and relax

Gap (2 min) Viewing slides of pictures pertaining to

mountains, rivers etc. with logical linking among pictures.

D1 (5 min) Play of logical lecture on cyclic meditation




Post (5 min) Sit and relax

D1-D4 are sessions of interventional ekagrata meditation state

D. Data Acquisition and Analysis

The lead-II ECG signal of each subject was recorded at

sampling frequency of 250 Hz using 4-channel RMS

polyrite ECG recorder (Chandigarh, India) in all sessions

as elaborated above. The acquired ECG data was visually

inspected and only visibly artifact free data were used.

The ECG signal was filtered using FIR band-pass (0.5

Hz-100 Hz) filter to remove any high frequency noise and

base wander, while retaining the diagnostically useful

frequency range. The R-peaks were detected using highly

efficient new wavelet transform function based QRS

detection algorithm [10]-[12]. This resulted into highly

accurate RR-tachogram (HRV signal), which leads to

evaluation of highly reliable HRV indices.

E. Heart Rate Variability Measurement

1) Spectral domain HRV indices using optimized AR

model

The heart rate variability evaluated using FFT with

improved windowing technique gives highly accurate

spectral HRV indices magnitudes as window bias caused

by windowing of data is removed in this algorithm. So

HRV indices with this technique have been taken as

reference. The AR modal has smooth PSD curve and easy

identification of central frequencies in spectral bands

corresponding to intrinsic oscillators. Moreover, the

spectral resolution in case of AR technique is mostly

affected by order of the model because time resolution is

independent of frequency resolution, so even a shorter

window length can be used. Further, the order of model

affects the spectral band (VLF, LF & HF) powers and

other HRV indices. So, a criterion is chosen for model

order selection, such that variance of the prediction error

is minimized. The Akaike’s FPE (finite prediction error)

criteria for model order selection has the disadvantage

that magnitude of HRV indices differs from those

evaluated with fast Fourier transform (FFT) technique

[13]. In the present study, this model order has been

optimized and evaluated on the basis of total prediction

error (TPE). It was observed experimentally, that if the

highest order of model is evaluated corresponding to total

prediction error (TPE) of +1% more than least square

error computed with FPE criteria or till the trend of least

square error starts increasing, then this model order

provides HRV indices which are closely same to those

evaluated with FFT technique, without sacrificing the

inherent advantages of AR model. However, if model

order is evaluated corresponding to more than this +1%

of least square error, then the HRV indices again starts

deviating from those evaluated with FFT [14]-[17].

The R-peaks were detected using highly efficient

detection algorithm and RR-interval tachogram was

formed. This RR-tachogram was resampled at 4 Hz and

HRV indices were evaluated using above elaborated

optimized autoregressive (AR) model for all four

meditation states in their pre-during-post sessions. Three

frequency bands were considered as per the task force

guidelines: very low frequency (VLF, 0.04 Hz), low

frequency (LF, 0.04Hz-0.15Hz) and high frequency (HF,

0.15Hz-0.40Hz). The spectral band powers were

evaluated by summation of power in each of these

frequency bands. The physiological processes attributed

to the VLF band are not well defined, but it is believed to

be associated with thermoregulatory mechanisms and

rennin-angiotensin variations. LF signifies sympathetic

tone and parasympathetic tone is related to the HF. The

VLF, LF and HF power components may be computed in

absolute (millisecond square) or normalized units (n.u.).

The LFn.u. HFn.u. indicates proportion of autonomic

control by sympathetic and parasympathetic branches of

autonomic nervous system [18]-[21]. The time domain

HRV measures of mean heart rate (HR (bpm)) was also

evaluated from RR- interval series in the present study for

further supporting the spectral based analysis. The mean

103


©2013 Engineering and Technology Publishing

heart rate indicates the heart rate dynamics in the

meditation state.

2) Statistical analysis

The group values of various HRV indices for all four

meditation states are summarized as mean S.D. in pre,

during and post sessions in Table II. Since, the recordings

were made on subsequent days, so pre-session values of

HRV indices differs slightly under all four states.

Therefore each pre-session HRV indices values have

been taken as baseline and their percentage change in

magnitude during ekagrata meditation state and in post-

sessions with pre-session values as baseline indicates the

effect of ekagrata meditation state. The significance of

change during and in post-sessions were determined by

one way ANOVA test with pre, during and post session

group values formulating group values. The ‘p-values’

and ‘Fisher F-values’ determined by one way ANOVA

test have been quoted in Table II.

TABLE II. HEART RATE VARIABILITY DYNAMICS IN PRE-AND DURING EKAGRATA MEDITATION

HRV

measure

aSubjects

(n)

bBaseline

(Pre-session)

cDuring ekagrata

meditation state Post-session F p

LF (ms2) 30 348.59 132.57 540.92 264.39 612.68 297.30 9.54 0.000

HF (ms2) 30 289.42 55.25 251.39 67.79 253.63 66.22 3.40 0.038

LF/HF ratio 30 1.89 1.69 2.57 1.22 2.96 1.64 3.75 0.027

LFn.u. 30 58.13 14.91 68.43 9.40 70.36 12.32 8.41 0.000

HFn.u. 30 41.87 14.91 31.57 9.40 29.64 12.32 8.41 0.000

Ptotal (ms2) 30 1010.72 448.59 1353.73 557.69 1298.03 510.70 3.94 0.023

HR (bpm) 30 77.47 7.45 75.52 6.93 75.94 6.43 0.65 0.522

LF = power in low frequency; HF = power in high frequency; ms2=absolute units; n.u.= normalized units; LF/HF ratio = sympatho-vagal

balance; Ptotal=total power; HR= mean heart rate. aHRV indices in terms of mean and standard deviation calculated for all thirty subjects (n=30) participating in the study in pre,during and post

sessions of ekagrata meditation state. bPre-session values were considered as baseline.

cValues were averaged over D1-D4 sessions of meditation states.

III. RESULTS

The typical effect of ekagrata meditation state on heart

rate variability dynamics is shown in Fig. 1 for one of the

representative subject in pre and during ekagrata

meditation state for one of the session (D4). The result for

session D4 has been shown as the subject will effectively

transit into ekagrata meditation by the session.

0 50 100 150 200500

600

700

800

900

1000

1100

Beat number

RR

-in

terv

al (

ms)

RR-interval series in pre-session

(a)

0 0.1 0.2 0.3 0.4 0.50

0.2

0.4

0.6

0.8

Frequency (Hz)

PS

D (

*1

03*

ms2

)

PSD in pre-session

(b)

0 50 100 150 200500

600

700

800

900

1000

1100

Beat number

RR

-in

terv

al(m

s)

RR-interval series during ekagrata

(c)

0 0.1 0.2 0.3 0.4 0.50

0.05

0.1

0.15

0.2

Frequency (Hz)

PS

D (

*1

03*

ms2

)

PSD during ekagrata

(d)

Figure 1. Heart rate variability of a representative subject in pre-and

during one of the ekagrata session (D4). The power spectral density (PSD) (Fig. 1(b) and 1(d)) in pre-and during ekagrata meditation state

are shown alongside with their with their corresponding RR-interval series (Fig. 1(a) and 1(c)). The RR-interval series and their PSD curve

are shown for session D4 such that the effect of meditation states is

clearly observable. Only 200 numbers of RR-intervals are shown in above figure for clear visibility, although these are more in any 5-

minutes duration session.

A. Comparison of HRV in Pre- and During Ekagrata

Meditation State

To determine the mean effect of ekagrata meditation

state on heart rate variability dynamics the mean value of

HRV indices during multiple sessions of D1, D2, D3 and

D4 have been calculated for all thirty (n=30) subjects.

The quantitative results for these thirty subjects (n=30)

showing effect of ekagrata on heart rate variability

dynamics are summarized in Table II. The results are

presented in terms of mean and standard deviation (for all

thirty subjects) in pre, during and post sessions. The

variations in frequency domain HRV indices in pre- and

during ekagrata states are either significant or very

significant. However the time domain indices of rMSSD

(ms) do not show the consistency in results.

104



B. Comparison of Ekagrata Meditation States with

Baseline

The HRV indices during and in post-sessions were

compared with their baseline (pre-session) values. The

indices were differently affected and their magnitude of

variations in comparison to the baseline is shown in Table

III in terms of percentage change with respect to baseline

value.

TABLE III. PERCENTAGE CHANGE IN HRV INDICES IN REFERENCE TO

BASELINE (PRE-SESSION) DURING EKAGRATA MEDITATION STATE AND

IN ITS POST-SESSION (N = 30)

Variables During ekagrata meditation Post-session

LF (ms2) 55.17 75.76

HF (ms2) -13.14 -12.36

LF/HF ratio 35.92 56.61

LFn.u. 18.92 21.02

HFn.u. -24.61 -29.22

Ptotal (ms2) 33.91 28.4

HR (bpm)

-2.50 -1.91

+ve value = increase in reference to pre-session; -ve value = decrease in

reference to pre-session *Percentage change was calculated for change in mean value (for thirty

subjects) of particular HRV indices during and in its post-session in

reference to baseline.

Compared to baseline, the percentage increase in LF

power was very significantly higher during ekagrata

(p<0.01) and this persists even in post-session. The HF

power is lowered very significantly (p<0.01) during

ekagrata and it remains lowered in post- ekagrata session.

The autonomic balance (LF/HF Ratio) is sympathetic

tone dominant during and even in post-ekagrata session.

The normalized LF and HF have shown the same trend as

the absolute LF and HF does. The total power (Ptatal

(ms2)) increases very significantly during ekagrata

(p<0.05). The mean heart rate (HR (bpm)) is lowered

during ekagrata and remains lowered in post-session.

IV. DISCUSSIONS

The meditation state of ekagrata used as an

intervention for the present study is one of the meditation

states towards the path of attaining final state of Samadhi

as per yoga literature. The results pertaining to heart rate

variability dynamics during this meditation state has

revealed interesting autonomic control dynamics as

observed in Table II. The quantitative variations in HRV

measures in reference to baseline during and in post-

sessions shown in Table III comprehend the effect of

interventions. It was observed that there is increased

power in LF during ekagrata, which indicates that

sympathetic tone is markedly enhanced under ekagrata.

The vagal tone is markedly reduced under ekagrata.

This is indicator of the fact that heart rate variability is

reduced under ekagrata. The sympatho-vagal balance

(LF/HF ratio) is sympathetic tone dominant, thus

autonomic control become more responsive to peripheral

variations i.e. meditation state of ekagrata. The

normalized indices of LF and HF show the percentage

contribution of sympathetic and vagal tones in autonomic

control. The reduced mean heart rate (HR (bpm)) in

ekagrata is similar to other studies undertaken previously.

The above trends of various heart rate variability

dynamics related measures, such as sympathetic tone

(LF), sympatho-vagal balance (LF/HF ratio), total

variability (Ptotal (ms2)), normalized LF, HF sustains

even in their post-sessions.

This study assessed effect of meditation state of

ekagrata on heart rate variability in selected middle aged

male volunteers, who were having training in meditation

practices. Inclusion of female subjects would have added

wider applicability of the study. However, the subjects

taking part in the present study were regular yoga

practitioners and taking pre-session values of each HRV

measure as baseline or control is well justified as the

trained yoga practitioner has the ability to be in particular

meditation state at a given time. Future studies may help

to elucidate effect of ‘Om’ meditation over other forms of

meditation and how the autonomic control is regulated.

V. CONCLUSIONS

This study reveals very interesting autonomic control

dynamics under intervention of ekagrata meditation state.

The sympathetic tone is increased, where as the vagal

tone is reduced under ekagrata. The increased sympatho-

vagal balance under this meditation state further supports

this fact. Thus, the heart rate variability is reduced under

ekagrata meditation state. The increased total spectral

power Ptotal (ms2) may be due to increased visceral

dynamics due to interventional meditation state.

ACKNOWLEDGMENT

The authors are highly thankful to Department of

Electronics & Communication Engineering and Director,

Dr. B.R. Ambedkar National Institute of Technology,

Jalandhar for providing all necessary help to carry out the

present work. The authors are also highly thankful to

Electrical Engineering Department and administration of

Institute of Technology Roorkee, India for their all

administrative and technical assistance in providing the

various facilities for this work. The help in clinical

aspects of Yoga and interpretation of results in this study

rendered by research and support staff of Yoga Research

Centre at Patanjali Yogpeeth, Haridwar, India is highly

appreciated.

REFERENCES

[1] K. Howorka, K. Pumprla, J. Heger, G. Thoma, H. Opavsky, and J.

Salinger, “Computerized assessment of autonomic influences of yoga using spectral analysis of heart rate variability,” in Proc. 1st

Regional Conference IEEE and 14th BMESI, 1995, pp. 1.61-1.62.

[2] P. Raghuraj, A. G. Ramakrishnan, H. R. Nagendra, and S. Telles, “Effect of two selected yogic breathing techniques on heart rate

variability,” Indian J Physiol Pharmacol, vol. 42, no. 4, pp. 467-472, 1998.

[3] C. K. Peng, J. E. Mietus, L. Yanhui, K. Gurucharan, et al,

“Exaggerated heart rate oscillations during two meditation techniques,” International Journal of Cardiology, vol. 70, no. 6,

pp. 101–107, 1999.

105



[4] C. K. Peng, J. E. Mietus, L. Yanhui, K. Gurucharan, et al., “Heart rate dynamics during three forms of meditation,” International

Journal of Cardiology, vol. 95, pp. 19–27, 2004.

[5] V. Shusterman and O. Barnea, “Sympathetic nervous system activity in stress and biofeedback relaxation,” IEEE Engineering

in Medicine and Biology Magzine, vol. 24, no. 2, pp. 52-57, 2005. [6] E. Jovanov, “On spectral analysis of heart rate variability during

very slow yogic breathing,” in Proc. 27th Annual International

Conference IEEE-EMBS Engineering in Medicine and Biological Society, 2005, pp. 2467-2470.

[7] S. Patil and T. Shirley, “Effects of two yoga based relaxation techniques on heart rate variability,” International Journal of

Stress Management, vol. 13, no. 4, pp. 1-16, 2006.

[8] T. Fred, A. F. H. David, H. John, and T. Melissa, et al, “Effect of transcendental meditation practice on brain functioning and stress

reactivity in college students,” International Journal of Psychophysiology vol. 71, no 2, pp. 160-176, 2008.

[9] R. K. Sunkaria, V. Kumar, and S.C. Saxena, “A comparative study

on spectral parameters of HRV in yogic and non-yogic practitioners,” International Journal of Medical Engineering and

Informatics, vol. 2, no. 1, pp. 1-14, 2010. [10] R. K. Sunkaria, S. C. Saxena, V. Kumar, and A. M. Singhal,

“Wavelet based R-peak detection for HRV studies,” Journal of

Medical Engineering and Technology, vol. 34, no. 2, pp. 108-115, 2010.

[11] S. C. Saxena, V. Kumar, and S. T. Hamde, “QRS detection using new wavelets,” Journal of Medical Engineering & Technology,

vol. 26, no. 1, pp. 7-15, 2002.

[12] D. Singh, “Analysis and interpretation of heart rate variability and blood pressure variability,” Ph.D. Thesis, Department of Electrical

Engineering, Indian Institute of Technology Roorkee, Uttaranchal, India, 2004.

[13] D. Singh, V. Kumar, S. C. Saxena and K. K. Deepak, “An

improved windowing technique for heart rate variability power spectrum estimation,” Journal of Medical Engineering &

Technology, vol. 29, no. 2, pp. 95-101, 2005. [14] R. K. Sunkaria, “Psycho-somatic effects on heart rate variability

dyanamics,” Ph.D. Thesis, Electrical Engineering Department,

Indian Institute of Technology, Roorkee, Uttrakhand, India, 2010. [15] J. L. A. Carvalho, A. F. Rocha, I. dos Santos, et.al, “Study on the

optimal order for the auto-regressive time frequency analysis of heart rate variability,” in Proc. 25th Annual International

Conference of IEEE- EMBS, 2003, pp. 2621-2625.

[16] R. W. De Boer, J. M. Karemaker and J. Strackee, “Relationship between short term blood-pressure fluctuations and heart rate

variability in resting subjects II: A simple approach,” Medical & Biomedical Engineering & Computing, vol. 23, no. 4, pp. 352-358,

July.1985.

[17] A. J. Camm, M. Malik, J. T. Bigger, G. Breithardt, et.al, “Heart rate variability-standards of measurement, physiological

interpretation and clinical use. Task Force of European Society of Cardiology and the North American Society of Pacing and

Electrophysiology” European Heart Journal, pp. 1043-1065, 1996.

[18] G. Baselli, D, Bolis, S. Ceruti, and C. Freschi, “Autoregressive modeling and power spectral estimate of R-R interval time series

in arrhythmic patients,” Computers and Biomedical Research, vol. 18, pp.510-530, 1985.

[19] P. Aurelien, R. Manuel, A. J. Sophie, B. Claire de, and D Andre,

“Spectral analysis of heart rate variability: interchangeability between autoregressive analysis and fast Fourier transform,”

Journal of Electrocardiology, vol. 39, no 9, pp. 31-37, 2006. [20] M. Malik and A. J. Camm, “Heart Rate Variability,” Armonk,

Futura Publishing Company Inc., New York，vol 118, no 6, pp.

436-447, Mar 1993. [21] J. Nayak, P. Subbanna Bhat, R. U. Acharya, U. C. Niranjan and O.

Wai Sing, “AR modeling of heart rate signals,” in Proc. IEEE TENCON, 2004, pp. 422- 426.

Ramesh Kumar Sunkaria was born in Kot Khalsa, Amritsar (Punjab),

on February 20, 1967. He received his B.Tech. degree in Electronics Engg from GNDU, Amritsar and M.Tech. (Hons.) degree in Electronics

and Communication Engg. from GNDEC, Ludhiana in 1990, and 2004

respectively. He received his Ph.D. degree from Electrical Engineering Department, IIT Roorkee in June, 2010 under the guidance of Dr. Vinod

Kumar, Professor and Head, Electrical Engineering Department, IIT Roorkee, Dr. Suresh C. Saxena, Director, IIT Roorkee and Dr. Achala

M. Singhal, Professor and Head, Cardiology Department, HIHT,

Dehradun (Uttrakhand). He has worked as Engineer in BHEL, Haridwar upto 1994 and Assistant Director (Engg.) in IBES, Govt. of India till

August 1996. At present he is Assistant Professor in Department of Electronics & Communication Engineering, Dr. B. R. Ambedkar

National Institute of Technology, where he has been teaching UG/PG

courses related to Biomedical Signal/Image Processing and Digital Signal Processing and Its Applications. His professional research

interests are in Medical Signal/Image Processing, Heart Rate Variability Signal Analysis and Digital Signal Processing.

106



Documents

Journal of Industrial and Intelligent Information Vol. 1, No. 2, June … · 2013-07-03 · cyclic meditation (CM) and supine rest (SR), showed decrease in LF power and LF/HF ratio