IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
1. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
Heart Pulse Monitoring: The Smart Phone Way
Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek
Deshmukh
(Department of Information Technology, Sinhgad College of Engineering, University of Pune, Pune)
ABSTRACT
It's never been easier to know what your condition and vital signs during that time will
heart rate is! This report is based on designing an enhance the responder's ability to assist them in the
cell phone application that will measure your best possible way. This report concentrates mainly on
heart rate or in simple words a Cardiograph. two vivid regions: one is the diagnosis of vital signs
Without any external hardware, just using a built- of the body and how a mobile phone can help in
in sensor of your Smartphone or tablet, you can biotelemetry, and the other is the means of sending
get accurate readings almost instantly. This all the data through remote connection in case of a
technique will enable the user measures your life threatening emergency situation. I have done
heart rate. You can save your results for future extensive studies in vital signs identification from a
reference, keep track of multiple people with person and propose new methods of using mobile
individual profiles, add notes and locations, and phone's sensors for quantifying the vital sign. These
even print out your measurements for sharing or methods utilize the accelerometer, video camera in
safe keeping. It uses your device's built-in sensors tandem with a LED ash and microphone sensor to
to calculate your heart's rhythm - the same detect and deliver a value for the needs. In this
approach used by professional medical chapter, I provide a general discussion about the vital
equipment. signs and the need for its diagnosis.
Knowing how fast your heart is beating
can be very useful when exercising, if you're 2. VITAL SIGNS
under stress, if out have a heart-related medical Vital signs are the most basic functions that
condition, or even just out of curiosity. Every can be measured from a person; they indicate their
measurement taken can be saved to your personal physical condition and wellness. When the
history, so you can keep track over time. In measurements tend to move away from normal, an
addition to the date and time of the measurement, abnormality in the physical status can be inferred.
you will be able to save the location at which it Most medical conditions can be diagnosed through
was taken (and see them on a map). This app can vital signs and confirmed with the help of special
be perfectly tailored to allow multiple people to tests. Each vital sign is measured differently with the
use the app on a shared device. A profile can be use of specialized equipments. These equipments are
created for each of your family members or not handy and do not come in miniature sizes for
friends, and each of them has their own individual portability. Hence I introduce the concept of
measurement history converting a mobile phone, which people use in their
day to day life into a vital sign diagnosing tool. There
1. INTRODUCTION are four vital signs which are standard in most
With the advent of internet, a lot has medical settings:
changed in people's lives. Users can stay in the house _ Pulse rate
and do almost any kind of activity like shopping, _ Respiratory rate
movies, entertainment, physical exercise, without _ Blood pressure
physically being at the appropriate place. In short, we _ Temperature
can tell that our lives are made simpler, faster and
efficient, if we leave behind the negative aspects of it. 2.1 Pulse rate (HR): Pulse rate is the rate at which
In the heart beats, measured either in the wrist or neck
The field of telecommunication, video given by beats per minute. The pulse rate is
phones have a wide scope in bringing people together influenced by the expansion of the arterial wall for
with face-to-face communication. Furthermore, video every beat. The most prominent spots for the pulses
transmission from mobile phone as a video call are wrist (Radial artery), neck(Carotid artery), inside
enhances the mobility of users. In case of a 911 of the elbow (Brachial artery), behind the
situation, this will allow people to make knee(Popliteal artery) and ankle joint (Posterior tibial
conversations a lot more understandable even under artery) [1]. The pulse rate varies with age and also
chaos. The responders could arrive at a decision more depends on the physical and psychological effects on
quickly, when viewing the live video feed of the the body. Higher pulse rate indicates the presence of
person. The caller's Information about the physical abnormality in the body and can also be caused by
other reasons such as anxiety, anger, excitement,
509 | P a g e
2. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
emotion, heart disorders, asthma, a large meal and so blood pressure of a person. Due to various reasons,
on. The pulse rate of an individual can help in the average blood pressure differs from each
determining various problems within the body, but it individual. The pressure values are categorized into
cannot be used solely to diagnose an abnormality. five major divisions. Table 1.2 shows the categories
Pulse rate is just a basic tool for diagnosis and hence of people in accordance to their blood pressure range.
can be used only for primary diagnosis.
Table 1.2. Categories of Blood Pressure
2.2 Respiratory rate (RR): Respiratory Rate is the Category Systolic Diastolic
number of breaths a person takes within a certain (mmHg) (mmHg)
amount of time or more formally, defined as the
number of chest movements involving inspiration
and expiration per unit time. The RR is measured in Hypotension <90 <60
units of breaths per minute. It is measured by Normal 90-120 60-80
counting the number of breaths (number of times the Pre- 121-139 or 81-89
chest rise) for a minute, usually when the person is at hypertension 140-159 or 90-99
rest. Respiratory rates will increase as the demand for Hypertension 1 >=160 or = 100
oxygen increases; it also increases due to illness, Hypertension 2
intensive physical activity, etc. The average RR
reported for a healthy adult at rest is usually given as
12 breaths per minute (12/60 Hz) [2] and the 2.4 Temperature (T): Temperature is one of the
estimates vary between 12-20 breaths per minute, other important vital signs. There is no direct way of
whereas the respiratory rate is higher in the case of measuring the person's temperature from the mobile
young adults, children and babies. As an individual device as of now. Mobile phone manufacturers have
age, breathing rate declines. In slow rates, more started incorporating onboard eco temperature
accurate readings are obtained by counting the sensors to mobile phones. It is just a matter of time
number of breaths over a full minute. Table 1.1 [3] before the temperature of surroundings and a human
shows the heart rate and respiratory rate at varying body can be measured using the phone.
ages showing a gradual decline in the rate with age.
3. MEANING OF PULSE:
Table 1.1. Heart Rate and Respiratory Rate for A pulse is a sudden burst of blood to the
Different Ages circulatory system when the walls of the heart
Age HeartRate RespiratoryRate contract. Heart rate or pulse rate is defined as the
(beats/min) (breaths/min) number of heart beats or pulses in a minute. The
Newborn 100-160 30-50 human heart comprises the atrium and the ventricles,
0-5 months 90-150 25-40 which coordinate to form a complete pumping action.
6-12 months 80-140 20-30 Approximately 2000 gallons of blood is pumped by
1-3 years 80-130 20-30 the heart every day. A Heart beat cycle consists of
3-5 years 80-120 20-30 two components, namely systole and diastole. Systole
6-10 years 70-110 15-30 occurs when there is an electrical impulse generated
11-14 years 60-105 12-20 by the Sinoatrial(SA) Node, triggering the heart to
14+ years 60-100 12-20 contract. Diastole occurs when the heart is relaxed.
Systole and diastole alternate each other to produce a
heartbeat. The heart rate is not just about how fast the
2.2 Blood pressure (BP): Blood Pressure is a force heart is beating; it is a regulatory mechanism for
exerted by blood on the walls of arteries, veins and delivering oxygen to the muscles to keep up the
the chambers of the heart. Blood pressure is one the demand.
most important vital signs and the body 3 maintain it
by interacting with the volume of blood and the force 4. MEDICAL WAY:
of contraction of the heart. During each heartbeat, BP Acoustically, the heart rate is measured by
varies between a maximum pressure called systolic listening to the heart beats, which are amplified
pressure and a minimum pressure called diastolic through the use of a stethoscope. Usually the
pressure. It is measured on the inside of an elbow at numbers of beats for a small interval of time, say 10
the brachial artery, which is the upper arm's major seconds, is observed and obtained for a minute by
blood vessel that carries blood away from the heart. multiplying with 6. In the same way, the pulse felt at
A person's BP is usually expressed in terms of the the wrist and neck can be measured and directly
systolic pressure and diastolic pressure values. An related to the heart rate. Figure 2.1 shows the regions
average healthy adult's pressure values read 120 where the pulse can be felt clearly for measurement.
mmHg during the systole and 80 mmHg during A more precise method of determining pulse rate
diastole. Pumping Rate, blood volume, resistance, involves the use of an electrocardiography (ECG or
viscosity, etc. are some of the factors which affect the EKG), pulse oximetry, etc. Shelley[4] discussed
510 | P a g e
3. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
about the effectiveness of pulse oximetry in the 5.1 Detail Working: The model works on the
detection of pulse even under noisy conditions where principle that, every heart beat pertains to a rush of
the use of stethoscope is hopeless. There are many blood in the blood vessels, even in the capillaries at
commercial heart rate monitors available in the the finger-tips. Whenever the capillaries are rich in
market which use two tiny blood during a systolic pulse, more light is getting
absorbed by the blood, leading to low
Figure 2.1. Prominent places for pulse detection.
Figure 5.2. Method of placing the finger over the
camera for heart rate measurement
Electrode strips to find the heart rate, the
same way an ECG works. These electrodes are
reactive index and darker frame intensities.
generally attached to some fitness gear or costume,
Likewise, during a diastolic pulse, most of the light
displaying the measurements on a screen. ECG uses
gets reflected leading to bright frames. This change in
the electrical activity of the heart over a period of
intensity of light which can pass through the finger
time, measured through the electrodes connected to
creates an alternative pattern of waves similar to a
the skin. These electrodes induce a tiny current of a
pulse. This change in intensity with time gives the
few µA into the body and detect electrical changes
heart rate of a person.
caused by the heart during each heart beat. These
changes are captured, amplified and delivered as an
output.
5. THE SMARTPHONE WAY:
Even with the presence of many
technologies for finding the heart rate of a person,
only a few of them are accurate to a certain degree. A
new model for heart rate estimation was proposed,
which worked on the concept of Photo
plethysmography (PPG), without using the
wavelength of light for analysis. Most mobile phones
in today's market come with a stock camera and
optionally a LED flash. These components were used
to define a system, deriving the heart rate of a person.
Figure 5.3. Architecture of heart rate system
In the proposed method, a video of short
duration was recorded, with the finger placed over
the lens of the mobile camera. The flash is turned
ON, so that adequate amount of light can reach the
Figure 5.1 shows the camera with flashlightin a
finger for proper measurement. For this experiment,
Smartphone
an application was developed for Nexus One[5] to
511 | P a g e
4. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
keep the LED flash consistently ON while recording
video from camera. The first three seconds of data
from the camera were discarded, since the CMOS
sensor of the camera tries to focus when turned ON.
Also, the camera doesn't need to be focused, as the
results rely only on the amount of light entering the
video feed. It was generally hard to detect the
fluctuations in the frames unless the pulses are
distinct. A similar methodology was used by
Banitsas[6] with a slightly different approach in the
analysis of video frames.
Figure 5.5. Filtered Data for analysis
Figure 5.4. Area under analysis
5.2 Implementation: Working of the system
comprises of six functional modules. Initially the
video frames were split into four quadrants and only
the first quadrant (Figure 5.4) was considered for
analysis, since most of the changes and fluctuations
are predominant in that region. Every pixel
information on each frame was split into individual (a)
Red(R), Blue(B) and Green(G) components. In most Figure 5.6. (a) Total Window of Data
samples, the prominent color applied only to R with
the others tending to zero in every frame, hence
difference in the red channel (Rc) intensity to that of
all the channels of a frame was negligible. For
accuracy of plots, only the Rc in video frames were
considered. The average intensity of pixels for every
frame was calculated as its frame intensity. The raw
intensity values were filtered with a moving average
filter to remove rough peaks from the graph for easier
identification of peaks. Figure 5.5 shows the filtered
results from the raw data obtained from finger pulse.
The entire frame was split into windows of fixed
length (Wt), for determination of peaks occurring at
equal intervals of time as seen in Figure 5.6. If the
pattern within Wt matched a sinusoidal pattern, the
heart rate was calculated by determining the number
of peaks (n) in the window and multiplying the peak
count with the window length as given by the
equation 1.
(b)
Figure 5.6. (b) Data spilt into smaller Time frames
512 | P a g e
5. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
Equation:1 window size. If the algorithm misses counting a peak
that should be present but
HR = n _ 60=Wt moved to the next window in a small fraction of time,
the obtained results could vary a lot. There are two
ways of calibration:
6. ACCURACY: (i) Average
This method showed encouraging results (ii) Window time calibration
with high percentage of accuracy. The collected data Of the two, measuring the average of windows is the
was validated with a commercial heart rate monitor easiest of the calibration methods. Two or more
available at a fitness center. To prove the windows could be taken and the average of number
effectiveness of the proposed system, higher heart of peaks could be computed to give a better result.
rate to the subject with excessive physical activity However, there is a small disadvantage in
was induced. From Table 6.1 it can be seen that the incorporating this method. If there is not enough
method gives a high percentage of accuracy from the legible data available for calculating peaks in
obtained data in finding the HR of the person. multiple windows, the results will be erratic. Hence
Table 6.1. Accuracy of results at varying heart window time calibration more suitable for this
rate for a single subject application was considered. Figure 7.1 shows a
5 sec Window 10 sec Window simulated heart pulse wave. The wave could be
different in case of irregular heartbeat and illness. In
Actual
this technique, it was assumed that the heart rate data
HR
Value Accuracy Value Accuracy set is a perfect sinusoidal wave with equal interval
between peaks. Based on my observations from the
data, an algorithm was proposed.1.In summary, given
102 108 94.11% 102 100.0% the frame intensity value of a fixed time window, the
108 96 88.89% 102 94.44% number of peaks
114 108 94.74% 114 100.0%
132 132 100.0% 132 100.0%
154 144 93.51% 150 97.40%
It can also be seen that, with the increase in
window size for analysis, the error propagation
decreases to a very minimum. For better accuracy of
the result, the user should hold the finger over the
camera lens for a longer time. Table 5.2 shows the
expected error in the system when varying the
window size. For less error in the results for
considered samples, the window size should be kept
large. Even more, the accuracy of the heart rate for a
full minute of data is precise with the actual heart rate
measured manually, with a 100% accuracy all the
time. Based on the need, it is possible to measure
every single heart beat with precision.
Table 6.2. Expected error from the system based
on window size
Window Size Expected Error Error %
(sec)
5 ±12 0 - 11.1
10 ±6 0 - 5.6 Figure 7.1. Calibration Of heart rate
15 ±4 0 - 3.7
20 ±3 0 - 2.8 Algorithm 1 Window time Calibration
30 ±2 0 - 1.8
7. WINDOW TIME CALLIBERATION
FOR ACCURACY
The obtained results look promising, but
there is a lot of error introduced due to the smaller
513 | P a g e
6. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
8. IMPACT ON HEART RATE WITH AGE
INPUT: Video Intensity for a Time window (T) & PHYSICAL INTENSITY
The approach uses observational data and
OUTPUT: Calibrated Heart Rate(CHR) does not require much computational analysis,
however a sensitivity analysis is very helpful in
N ← no of peaks in Window /*N x 60/TGives predicting the outcome of the decision based on the
uncalibrated HR value*/ situational parameters. In the heart rate measurement,
the key parameters involved in the determination of
pk2pk ← AvgPeaktoPeaktime /*Timebetweentwo heart are the age and the intensity of work being
consecutive peaks*/ done. Even though there are some more biological
factors involved with it, they cannot be quantified at
Ex ← T – N x pk2pk / *Ex is non considered time this point. Researchers from Oakland University[7]
for Datapoints*/ have predicted the maximum heart rate of people by
age based on records collected over 25 years, giving
if Ex >= pk2pk then rise to a non-linear equation
N ←N+1
end if Table 8.1 Calibration results for Heart rate
if Ex >= pk2pk / 2 then
N ← N – (pk2pk - Ex ) x 60 / T
end if
CHR ← N x 60/ T
for the window are counted and the time
taken between consecutive peaks is calculated. Then
the average value for peak to peak time (pk2pk)
computed. If the occurrence of the peaks is harmonic,
the average will be same as an individual consecutive
peak time. This gives the time taken for one complete
cycle of systole and diastole. The wave marked in red
in Figure 2.7 is the region that is neglected during the
calculation of heart rate. Hence the total time of data
which is not being considered is denoted as Ex.
When Ex exceeds the time taken for a complete
cycle, one more peak is added to the calculated
number of peaks and the difference between Ex and
the cycle time is taken as the new unconsidered
value. If Ex exceeds half the cycle time, calculate the
fractional calibration time value for the peaks,
otherwise the number of peaks is taken as the heart
rate. The calibrated heart rate values showed greater
accuracy when compared to the un-calibrated data.
Commercial heart rate monitors had the same amount
of error induced due to time window computation. HR is the exact heart rate estimated for 60
Hence, the HR for a full 60 sec or 2 x 30 is seconds in a dataset; C - Calibrated result; UC – Un-
considered as the heart rate of a person for calibrated result; Acc.% is accuracy of Calibrated
comparison with the calibrated results of smaller data; Imp.% is improvement of accuracy in calibrated
window sizes. A single user's data set, taken at data from un-calibrated data for determining the
different time of the day, calibrated for different HRmax .
window sizes is shown in Table 8.1. From Table 8.1,
significant improvement in the results can be seen, Equation 2
much closer to the actual heart rate. Most of the data HRmax = 191:5 - (0:007 x age2)
items had improved accuracy, except a few marked in
red in the table which show decreased accuracy from
the actual result. This happens due o the irregular Gellish[8], proved the predicted heart rate
heart cycle, where the peak to peak interval varies lies in a tight range between ± 2 - 5bpm for average
largely to yield errors in calibration. The longer the individuals. However, the values vary within a wider
data gets, the results will be more accurate. With range for athletes. The actual
short data lengths, the chances for error propagation
will be high.
514 | P a g e
7. Mayur Lunawat, Aquibjaved Momin, Varun Nirantar, Abhishek Deshmukh / International
Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.509-515
[2] T. Gerard and A. Nicholas, Principles of
Anatomy and Physiology, Harper-Collins,
New York, sixth edition, 1990.
[3] Wikipedia, “Variation of vital signs with
age", http://en.wikipedia.org/wiki/Vital
signs.
[4] S. Kirk, “Photoplethysmography:Beyond the
Calculation of Arterial Oxygen Saturation
and Heart Rate", Anesthesia & Analgesia,
vol. 105, no. 6S Suppl, pp. S31-S36,
December 2007.
[5] Google Inc., “Nexus one: Technical
speci_cations", http://www.google.com/
phone/static/en US-nexusone tech
specs.html.
[6] K. Banitsas, P. Pelegris, T. Orbach, D.
Cavouras, K. Sidiropoulos, and S.
Figure 8.1. Sensitivity analysis for 10 to 80% Kostopoulos, “A simple algorithm to
increase in Age and Work Intensity monitor hr for real time treatment
applications", in Information Technology
heart rate depends on the intensity of work and Applications in Biomedicine, 2009.
involved by the person. Resting heart rate is the ITAB 2009. 9th International Conference
lowest heart rate, which can be achieved by a normal on, 4-7 2009, pp. 1 -5.
healthy person. Even at rest, the intensity of work [7] “Medicine & science in sports & exercise",
will be a little more than 25 % and the heart rate The Official Journal of the American
increases as the intensity of work increases. There College of Sports Medicine, vol. 39, no. 5,
exists a direct relationship between the heart rate and pp. 749-898, 2007.
the work intensity, hence the percentage a work [8] G. Ronald, G. Brian, O. Ronald, M. Audry,
intensity for age gives the heart rate at that moment. R. Gary, and M. Virinder, “Longitudinal
From Figure 8.1, it can be seen that the HR gradually modeling of the relationship between age
decreases with age and increases with work intensity. and maximal heart rate", Medicine &
The resting heart rate of people can go well below Science in Sports & Exercise, vol. 39, no. 5,
normal and reach 45-60 bpm [9] for normal pp. 503-508, 2007.
individuals. The results of sensitivity analysis show [9] “What should you know about your heart
the normal range of heart rate which a person can rate or pulse", http://www.nemahealth.org/
achieve based on the age and work percentage. It is programs/healthcare/heart rate pulse.htm.
also evident that, the proposed methodology follows
the pattern and most of the values lie within the range
of the graph, proving its effectiveness.
9. INFERENCE
This chapter we have discussed over the
normal medical protocol and described a new system
that uses the camera on a mobile phone to find heart
rate. From the results, it was inferred that for longer
duration of data collection, there was a better chance
of achieving more accurate heart rate. It was also
observed that the accuracy could be improved from
87% with a 5 sec data to 99 % with a 30 sec data
without any calibration. After using the calibration
method, it was found that there was an improvement
in the accuracy up to 5 or 6 % for even small sized
data. This method worked well for 15 sec data and
helped achieve 98 % accuracy most of the time.
REFERENCES
[1] N. Abhijit, “Normal pulse rate",
http://www.buzzle.com /articles/normal-
pulserate. html.
515 | P a g e