You are on page 1of 5

Pulse Rate Variability for Detection of Autonomic

Tone of an Individual
Suma P Dr. Suma K V
Dept. of Electronics and Communication Dept. of Electronics and Communication
Ramaiah Institute of Technology Ramaiah Institute of Technology
Bangalore, India Bangalore, India
sumap1994@gmail.com sumakv@msrit.edu

Abstract— Excessive psychological stress has major impact on the IR light coming from the LED and the photosensor
Autonomic nervous system (ANS). Perceived stress is one of receives the light from the skin tissue which is placed on the
the outcome caused due to stressful audio-visual stimulus that other side.
makes the person feel threatened or upset. Pulse rate
variability (PRV) is an approach which detects the variation in
ANS caused due to stressful stimulus. To compute PRV,
Photoplethysmography (PPG) signal was acquired using
fingertip pulse sensor which is interfaced to a microcontroller.
From PPG signal, PRV parameters are computed in time
domain, frequency domain and geometric domain which helps
to quantify the level of stress induced due to stressful audio-
visual stimulus. These parameters are then transmitted to the
server through GSM module. The PRV parameters are
retrieved from server and displayed on an Android app which
is easier to monitor the stress anywhere and anytime.

Keywords—Photoplethysmography, Pulse rate variability,


Autonomic nervous system, Audio-Visual stimulus

I. INTRODUCTION
Psychological stress is one of the major mental health
problems of modern society. It is related to many mental
health issues such as anxiety and depression disorders. So
the purpose of this study is to investigate the effect of
Perceived stress on Autonomic Nervous System (ANS) due
Fig. 1. Framework of the proposed PPG monitoring system
to stressful audio-visual stimulus. Several parameters like
electroencephalograph, pulse rate variability, blood pressure
The firmware computes pulse rate expressed in Beats per
among others quickly respond to any change that occurs in
minute (BPM) and interval between adjacent PPG pulses
mental stress level. Pulse rate variability is one of the most
called as RR interval is obtained. After determining RR
reliable indicator of stress and Photoplethysmography is a
intervals the time domain, frequency domain and geometric
simple and low-cost optical technique which is efficient in
domain analysis is computed. This analysis indicates overall
computation of PRV [1]. PRV has the unique character of
balance between the sympathetic and parasympathetic
measuring the condition of Autonomous Nervous System.
systems.
The autonomic nervous system is the part of the nervous
system which is responsible for controlling involuntary II. LITERATURE SURVEY
functions such as heartbeat, blood flow and has two main
divisions: Sympathetic and Parasympathetic system. R. Logier et al. [1] describes the PRV analysis that can be
Generally, the sympathetic division prepares the body for used as an alternative to heart rate variability (HRV) for
stressful situations and it increases heart rate and expands portable systems available on the market. The correlations
the airways which helps breathing easier. The between HRV and PRV parameters in time domain and
parasympathetic division controls body process during frequency domain are 0.96. This enables us to establish the
ordinary situations and therefore it decreases the heart rate fact that there is a link between HRV and PRV parameters.
and restores body to normal state. As shown in Fig. 1, PPG
signal is acquired using pulse sensor and it is based on Mohammad Ghamari et al. [2] incorportes a wireless device
photoplethysmography technique which detects the change to monitor the heart rate based on PPG technique. An
in blood volume using light source and detector. It is used Android app was also used to enable the smart device to
non-invasively to measure at the surface of the skin. Pulse display the received PPG signal. A Matlab program was
sensor consists of an IR LED and a photosensor which are used to analyse and detect the R-R peak intervals of the PPG
placed on two opposite sides and are facing each other. waveform. Finally by feeding the PPG signal peaks into the
When a sensor is placed on the fingertip, it is illuminated by Kubios software, the HRV analysis report was generated.
P Madhan Mohan et al. [3] works on HRV that is
determined by using Photoplethysmographic (PPG) optical
sensor. During this process the stress level of the subject is
assessed from the HRV data. Data is collected from persons
in rest position also with variable breathing rates in order to
simulate stress. LF/HF ratio is calculated which is a measure
of stress of a person.

III. SYSTEM DESCRIPTION


Fig. 3.The block diagram of PPG acquisition system
The PPG monitoring system as shown in Fig. 2 comprises of
PPG signal acquisition unit and display unit. B. Display Unit
In this work ThingSpeak server which is an open-source
Internet of Thing application is used to store and retrieve
data from GSM module. Thingspeak server has the
capability to display the waveforms of the parameters
computed in real time. From the server the data is retrieved
on an android app to display the values in real time as
shown in Fig. 4. Also LCD module is interfaced with
microcontroller to display the PRV parameters as shown in
Fig. 5.So the LCD module is interfaced with microcontroller
in 4 bit mode. The PRV parameters displayed on 16x2 LCD
indicates P for pulse rate, R for RRinterval, F for LF/HF
ratio, S for SDNN and SD for SD2/SD1 ratio.

Fig. 2.The block diagram of PPG acquisition system

A. PPG Signal Acquisition unit

As shown in Fig.3 Pulse signal is acquired by interfacing


Fig. 4. Display of PRV parameters in an Android App
pulse sensor with microcontroller [4]. The sensor produces
analog signal so it is connected to pin A0 of the
microcontroller. As microcontroller has an inbuilt ADC so it
converts the analog signal from pulse sensor to digital
signal. After acquisition of pulse signal, PRV analysis is
computed in time domain, frequency domain and geometric
domain. These parameters are then transmitted to GSM
module. SIM800C GSM module is used to receive the PRV
parameters from microcontroller and transmits these Fig. 5. Display of PRV parameters in 16x2 LCD
parameters onto the server.
IV. METHODS PPG signal is acquired from pulse sensor. The interval
between adjacent PPG signal called as RR Interval is
A. Participants determined. Then Time-domain, Frequency domain and
35 subjects, apparently healthy males and females, aged Geometric domain parameters adopted in PRV analysis
from 18 to 24, took part in this study. Before the experiment were extracted from RR Interval in PPG recording as
general information about age, gender, height and weight of indicated in Fig. 7 [8].
the subject was considered. Then a perceived stress level In time domain analysis SDNN and RMSSD was computed.
questionnaire form was distributed to the participants to SDNN is the estimate of overall PRV and examine the
check the level of stress the individual has experienced in ability of the heart to respond to stress. SDNN values
the previous months. The subjects were asked to be free predict unhealthy conditions. Below 20ms is an indication
from coffee and alcohol drinks atleast two hours before each of chronic stress related disease.
experiment.
B. Procedures SDNN is computed as indicated in equation 1:
All subjects were tested during normal and stimulus session. 1
SDNN = √ ∑ni=1(RR i − ̅̅̅̅
RR)2 (1)
The test procedure is shown in Fig. 6. Prior to the start of n−1
the experiment the subject was asked to be in the rest
position for 15 min as indicated from a to b. During normal where 𝑅𝑅𝑖 denotes the time from the ith to the (i+1)st Rpeak
condition, PPG readings was recorded for duration of 5min and ̅̅̅̅
𝑅𝑅 is the average interval, giving n intervals in total.
as indicated from b to c. Then a break of 5min is provided
before applying the stimulus. Then a stressful video with RMSSD estimate high-frequency variations in pulse rate in
sudden frightening sound is applied for duration of 5min as short term RR recordings. Decrease in RMSSD below 10ms
shown from d to e to induce stressful stimulus and to elicit a results in development of cardiac disease.
fear response from the subject which is associated with the
psychological change which has an impact on PRV[5][6]. RMSSD is computed as shown in equation 2:
After 5 minutes of stressful stimulus 15min of recovery 1
period was considered from e to f to check exactly the time RMSSD = √ ∑𝑁−1
𝑖=1 (𝑅𝑅𝑖+1 − 𝑅𝑅𝑖 )
2 (2)
𝑁−1
at which the individual recovers to the normal state.
Therefore when stimulus is applied there is decrease in the Frequency domain analysis include LF/HF ratio which is a
value of time domain and increase in geometric domain measure of overall balance between sympathetic and
parameters when compared to that of normal condition [7]. parasympathetic systems. LF is computed between 0.04 to
It is an indication of presence of mental stress caused by 0.15Hz and HF is computed between 0.15 to 0.40Hz.
audio-visual stimulus which is verified by perceived stress It is computed as shown in equation 3:
level questionnaire which is given to an individual before
playing audio-visual stimulus.
0.15
𝐿𝐹 ∫0.04 𝑓(λ) 𝑑λ
𝐻𝐹
= 0.40 (3)
∫0.15 𝑓(λ) 𝑑λ

Where f (λ) is determined from RR interval series that is


f(λ) = (120* RRInterval + 660)
Geometric domain analysis includes the ratio of standard
deviation SD2 and SD1 which reflects sympathetic to
parasympathetic activity. It is computed as per the equations
4 to equation 7.

𝑆𝐷1 = √𝑉𝑎𝑟(𝑥1 ) (4)


Fig. 6. Flowchart of the experiment
𝑆𝐷2 = √𝑉𝑎𝑟(𝑥2 ) (5)
C. PRV Parameters:
Var (𝑥1 ) and Var (𝑥2 ) is the variance of 𝑥1 and 𝑥2 .
Table I indicates definition of PRV parameters.
Parameter Unit Units Description
𝑥1 and 𝑥2 is calculated as shown in equation 6 and 7:
Interval between
RR INTERVAL millisecond consecutive normal R
𝑅𝑅𝑖 +𝑅𝑅𝑖+1
waves in PPG waveform 𝑥1 = (6)
√2
Standard deviation of the
SDNN millisecond 𝑅𝑅𝑖 − 𝑅𝑅𝑖+1
RR intervals 𝑥2 = (7)
Root mean square of √2
RMSSD millisecond successive differences in
RR intervals
Ratio of Low frequency to
LF/HF ratio
High frequency
SD2/SD1 ratio Ratio of Standard deviation
From RR interval, PRV analysis is computed in time domain,
frequency domain and geometric domain for the required
interval of RR series. The Table II indicates the PRV
parameters computed for different subjects and all the values
are within the standard range. The above 5 subjects are
considered at normal condition in which all the parameters is
in the standard range which indicates increase in pulse rate
variability which is an indication of healthy condition.
B. VALIDATION OF PPG SIGNAL
PPG signal acquired from pulse sensor is validated by
comparing with ECG signal. ECG signal analysis is
performed using RMS Vagus SoftwareTM. PPG signal
analysis is performed in arduino software. PPG and ECG
readings were recorded simultaneously for a duration of 5
min from the subject. A close accuracy is achieved with
average percentage error of 8% for BPM and 3% for RR
interval when compared with that of ECG signal as
indicated in Table III.
C. COMPARISON OF STRESS LEVELS DURING PRE AND POST
STIMULUS CONDITIONS FOR DIFFERENT SUBJECTS
When audio visual stimulus is applied the mean RR interval,
SDNN, RMSSD was significantly reduced and SD2/SD1
ratio was significantly increased when compared with that
of normal condition. The frequency domain parameters
LF/HF ratio showed no significant difference as indicated in
the Table IV. The short term PRV measures are computed in
the presence and absence of acute audio visual stress. When
stimulus is applied it stimulates sympathetic system which
reflects significant decrease in RR interval, SDNN, RMSSD
and significant increase in SD2/SD1 ratio.

The lowered mean RR interval indicates a higher resting


pulse rate during the stimulus. It is an indicator of
Fig. 7. Flowchart of the proposed system sympatho-vagal balance. In normal condition sympathetic
activity is less which results in increase in the value of RR
V. RESULTS AND DISCUSSION interval. When stimulus is applied, sympathetic activity is
` high which results in decrease in the value of RR interval.
A. PULSE SENSOR OUTPUT
The significant decrease in SDNN and RMSSD in the
Initially a fingertip pulse sensor is used to acquire pulse presence of stimulus reflects a decreased total PRV and it is
signal from which pulse rate in BPM and the adjacent an indication of influence of chronic stress. In normal
interval between PPG pulses called as RR interval also condition there is an increase in SDNN and RMSSD values
known as Inter Beat Interval (IBI) is computed which is as which results in higher PRV.
shown in Fig. 8. The significant increase in SD2/SD1 ratio in the presence of
audio-visual stimulus is an indication of increase in
sympathetic activity which results in increase in stress level
whereas in normal condition is an indication of decrease in
SD2/SD1 ratio.

Therefore the presence of mental stress caused by audio-


visual stimulus is verified by perceived stress level (PSS)
questionnaire which is given to an individual before playing
audio-visual stimulus. It helps to classify between average
stress level and high stress level. If the stress level is within
the range from 13-19 it is classified as average stress level,
if the stress level is greater than 20 then it is classified as
high stress level. Also after applying the stimulus it is found
that the person having high stress level requires more time
Fig. 8. Acquisition of PPG signal from pulse sensor to recovery to the normal state.
Table II. Sample PPG readings for 5 different subjects

Parameters Standard range Subject-1 Subject-2 Subject-3 Subject-4 Subject-5


BPM 60-100 BPM 69 72 67 71 100
RR Interval (ms) 500-1000 ms 788 694 812 804 524
SDNN (ms) 32-93 ms 65.72 64.08 84.74 64.56 60.02
RMSSD (ms) 19-75ms 47.04 63.44 62.45 59.15 69.81
LF/HF ratio Table II. Sample PPG
0.1-8.7 readings for 51.03
1.02 different subjects
1.02 1.02 1.05
SD2/SD1 ratio 0.1-4.5 1.82 2.69 3.53 2.47 2.34

Table III. Mean values of PPG and ECG signal for 10 subjects

Computed from PPG Computed from ECG


Parameters Table (mean
II. Sample
± SD) PPG readings for 5(mean
different
± SD)subjects
BPM 82.6 ± 3.43 89 ± 5.43
RR Interval (ms) 642 ± 35.18 677.28 ± 35.91
SDNN (ms) 38.55 ± 9.97 37.70 ± 6.61
RMSSD (ms) 26.98 ± 8.61 20.36 ± 6.62
LF/HF ratio 0.97 ± 0.005 2.19 ± 1.48
SD2/SD1 ratio 2.69 ± 0.49 3.72 ± 0.73

Table IV. Mean values of PRV parameters for 35 subjects

Normal Condition Audio-Visual Stimulus


Parameters Ground Truth
(mean ± SD) (mean ± SD)
BPM 75.66 ± 8.20 76.66 ± 9.78
RR Interval 724.26 ± 94.53 715.33 ± 99.65
SDNN 57.24 ± 9.62 42.16 ± 9.44 As per Perceived Stress
RMSSD 49.55 ± 14.55 35.44 ± 10.85 Level Questionnaire
LF/HF ratio 1.03 ± 0.016 1.04 ± 0.017
SD2/SD1 ratio 2.36 ± 0.47 2.60 ± 0.42

VI. CONCLUSION REFERENCES

[1] R. Logier, J. De Jonckheere, A. Dassonneville, M. Jeanne,"Comparison


In this modern era youths spend more time in front of of pulse rate variability and heart rate variability for high frequency
television in which they are influenced by stressful stimulus content estimation", 2016 38th Annual International Conference of the
which will have an adverse effect on their health. In order to IEEE Engineering in Medicine and Biology Society (EMBC), pp. 936-
evaluate the impact of stressful stimulus on humans a non- 939, 18 October 2016.
invasive PPG monitoring system has been designed and [2] Mohammad Ghamari, Christopher Aguilar, C. Soltanpur, Homer
implemented to extract PRV parameters and a series of Nazeran, "Rapid Prototyping of a Smart Device-Based Wireless
experiments have been performed to determine the changes Reflectance Photoplethysmograph", 2016 32nd Southern Biomedical
in PRV features under normal condition and during audio- Engineering Conference (SBEC), pp. 175-176, 28 April 2016.
visual stimulus condition. The result that is obtained [3] P Madhan Mohan, V Nagarajan, Sounak Ranjan Das,"Stress
indicates that the PRV parameters changed significantly measurement from wearable photoplethysmographic sensor using heart
with increase in sympathetic activity during stressful rate variability data", 2016 International Conference on
stimulus condition which reflects in decrease in time domain Communication and Signal Processing (ICCSP), pp. 1141-1144, 24
November 2016.
parameters and increase in geometric domain parameter
when compared with that of normal condition. So this study [4] Ravim, Suma K V, “Hardware simulation for bio-signal acquisition
aims to create awareness to the entertainment industry to system”, 2016 International conference on Applied Science,
release the videos such that the videos should not cause Engineering and Technology (ICASET-16), 6-7 January 2016.
mental stress on humans on watching those videos. [5] Kun-chan Lan, Paweeya Raknim, Wei-Fong Kao, Jyh-How Huang,
This system can be used to diagnose mental disorders in “Toward Hypertension Prediction Based on PPG-Derived HRV
rehabilitation centres. This helps in providing relevant signals”, Journal of Medical Systems 42, no. 6 (2018): 103.
timely treatment. It is a very powerful tool in the field of
[6] Jing-Jhao Ye, Wan-Chun Lin, Kuan-Ting Lee, Yu-Ting Tai,
psychiatry research for studying various psychiatric “Photoplethysmography variability as an alternative approach to obtain
disorders. heart rate variability information in chronic pain patient”, Journal of
ACKNOWLEDGMENT Clinical Monitoring and Computing 29, no. 6 (2015): 801–806.

[7] Mimma Nardelli, Gaetano Valenza, Alberto Greco, Antonio Lanata and
The authors would like to express their gratitude and respect Enzo Pasquale Scilingo, “Recognizing Emotions Induced by Affective
towards Dr. Venkatesh D and Dr. Arun Kumar, Department Sounds through Heart Rate Variability”, IEEE Transactions on
of Physiology, Ramaiah Medical College, for timely advice, Affective Computing 6, no. 4 (2015): 385-394.
suggestions and regular assistance in our project work and
[8] Marcus Vollmer, "A robust, simple and reliable measure of heart rate
also to Ramaiah Institute of Technology for the logistical variability using relative RR intervals", 2015 Computing in Cardiology
support provided during the completion of the study. Conference (CinC), pp. 609-612, 6 September 2015.

You might also like