Professional Documents
Culture Documents
Abstract:
Remote patient monitoring systems are based on a combination of advanced remote monitoring devices,
telecommunication technology, and innovative software and hardware. As mobile phones become more
powerful and perform more complex interactions between mobile devices to resident software and other
server based software, they have been recognized as effective tools for telemedicine. So, the merging
of the Internet and mobile computing, introduce new opportunities and challenges in the telemedicine
sector. This study describes the development and implementation of an Android based smart phone in the
home monitoring health care system. The system utilizes Android devices as mobile access terminals
for patient-monitoring services. The main goal is subsequently divided into two parts: in the first part
we establish a reliable connection between mobile device and a sensor to collect continuous data from
patients. In the second part we provide ElectroCardiGram (ECG) andPhoto-PlethysmoGraphy(PPG) signal
filtering and processing, feature extraction, detection of any abnormalities in ECG and calculating heart
rate using the most familiar and multi-purpose MATLAB software.
Keywords:
Telemedicine Sector; Remote Patient Monitoring Systems; Android Smart Phone; Vital Signs Processing;
Patient Monitoring and Diagnosis
1. INTRODUCTION
Recently, healthcare for patients suffering from chronic diseases or even elderly or lonely people has been an
important research topic. The increase in life expectancy due to improvements in living standards, and medical
treatments, has resulted in aging population diseases in the last few years [1]. Therefore, the modern health caresystem
aims to enhance the safety and comfort of the patientslife while managing chronic diseases. This creates theneed to
develop home healthcare system, which will improve the patient monitoring devices and make them more mobile.
Therapid development of informationand telecommunication technologies has broughtgreat revolutions in that field
[2]. Remote patient monitoring benefits not only the patients who are able to receive health care more efficiently, it
also benefits the doctors who can streamline their efforts to assist more patients.
In healthcare, There are many promising systems for remote physiological signals monitoring and diagnosis have
been developed. all the systems developed and currently used can be categorized by several areas: medical sensors,
data communication, monitoring devices and medical signal processing algorithms [3]. In the area of medical sensors,
research groups have mainly focused on developing sensor devices to be small size [4], maintaining patient mobility
36
[5] and consume low operating power [6]. A collection of medical sensors were designed for noninvasiveremote
monitoring of the following physiological signals:electrocardiographic signals (ECG) [7, 8], blood pressure [9, 10],
or all patients vital signs [11]. In the area of data communication, the medical staff can get patients physiological
data by connecting medical sensors to other networks such as cellular network or internet [3]. For real-time operation,
several monitoring systems transfers patients medical data through General Packet Radio Service (GPRS) networks
to medical center or hospital [12]. This can be costly when GPRS is used for continuous data transmission. To deal
with this issue other systems consider processing the medical data on a local device before transmission[13, 14]. In
these systems, the raw patients vital signsare transferred to the medical center through internet and only abnormal
cases are transferred through GPRS networks.In the area of medical signal processing and diagnosis, many medical
algorithms were developed in recent telemedicine studies to help in patient diagnosis [15] and early detection of
cardiovascular diseases [16]. Many investigations have proposed monitoring systems that can measure various
bio-signals and provide QRS detection and arrhythmia classification [16], real time ECG classification algorithm
[15], heart rate variability measurement [16].
New features of mobile devices(smart-phones, tablets), create new opportunities to usethem as devices for collecting
clinical data, delivering information to medical staff, researchers and patients. Patients vital signs are monitored in
real time and directly provide health care through mobile telemedicine. the usage of smart phone in patient monitoring
can be classified in two groups:The first group uses smart phones personal digital assistants (PDAs) equipped with
biosensors that record patients vital signs and transmit them to a healthcare center or hospital for analysis [17, 18].
Some solutions can store the signals locally as well [19].Most are capable of recording, viewing and processing ECGs
directly on the smart phone[20]. In [18], Leijdekkers and et al. developed remote healthcare monitoring system
which includes a smart phone, wireless sensors, web servers and IP webcams. The smart phone acts as a receiver
that receives all the data sent by the wireless healthcare sensor and further process the data by sending the data the
healthcare server or calls an ambulance during emergency. The TELEMON remote monitoring system allows patients
suffering from chronic disease to be monitored with a wireless body area network (WBAN) of sensors connected
by radio to a patient personal server (PDA) [12]. The PDA collects the and forwards it to the database server via
one of the WiFi or GSM/GPRS connection. The results of data processing are locally generated alarms displayed
by the software running on the central server. In [17], another system shows us an ECG monitor system based on
android smart phone. ECG signal is transmitted to an android phone and then be forwarded to a remote server.
Using a PC, doctors can view the ECG after logging in the server. The second group aims at building platforms for
real-time remote health monitoring. The platform developed by this second group collects the bio data and sends
it to a care center or a hospital for processing and analysis then showing them on doctors smart phone [21, 22].
Many researchers have done some studies on the application of smart phone in ECG monitoring and processing,
such as [17, 23]. MOLEC [14],provides a solution that analyses the ECG locally on a PDA. It generates alarms to
the hospital in case of high risk arrhythmias. Also There are some researchers proposed the way of alarming, most
of them use the short message service (SMS) or calls such as [23, 24]. In [24], pulse alarms details are described
in a SMS. But for an ECG image, it is hard to describe in a SMS. In [23], author use the multimedia messaging
service(MMS) and email to send the abnormal ECG images. The disadvantage of this way is that MMS is not able to
carry a picture in high resolution, and if using email as the transmitting way, doctors may not see the email timely,
and it also causes inconvenience. Our objective is to investigate and develop a healthcare system for continuous
patient monitoring and diagnosis. With this system the physiological parameters such as electrocardiogram, heart
rate, oxygen saturation (SPO2), PPG and temperature are obtained using mobile care unit attached to the patients
body then transmitted to a remote server via GSM/GPRS network. The key role of remote server here is to process
data in real time and distribute them on the caregivers smart phone. It receives ECG and other parameters data
from the mobile care unit; then filters the data, extract ECG features such as QRS complex, calculates heart rate,
arrhythmia classification and identifies abnormal ECG data using an intelligent data analysis scheme based on a new
cosine similarity measure.Finally patients vital signs and results of ECG analysis will be routed to doctors mobile
37
phone via GPRS network and will be displayed using friendly Android application in real time mode.
Heart rate
QRS duration
P wave
Bradycardia
Normal
Tachycardia
Normal
Ratio 1:1
P-R interval
Normal
38
The complete work is explained in the subsequent sections. Section 3 describes the architecture of the proposed
system including sensor part, server part and mobile part. The details of ECG and PPG filtering, QRS complex
detection algorithm and detection of arrhythmias their characteristics and methodology used for their detection are
explained in sections 4.2.1, 4.2.2 and 4.2.4.The proposed similarity measure for exploring abnormal ECG signal is
discussed in section 4.2.3. And the summary is concluded in section 5.
3. SYSTEM STRUCTURE
The mainobjective of this paper is to design and implement a telemedicine system with intelligent vital signs
analysis. It is based on physiological sensors, wireless communication, World Wide Web and smart phones for vital
signs monitoring, patient diagnosis and homecare. The architecture of the proposed system is shown in Figure 2.
Itmainly consistsofthe following parts.
4. SYSTEM COMPONENTS
This section details the system components of the proposed Android based telemedicine system for patient
monitoring and diagnosis.
39
is based on the ATmega328 microcontroller as the MCU of the medical care unit. It has input-output circuitry and
peripherals built-in, allowing it to interface more or less directly with real-world devices such as sensors. The main
functions of MCU in the proposed system are:
Processing the received signals using different sorts of processing techniques like filtering and digitizing the
sensor data.
Figure 4. Arduino UNO board connected to ECG, temperature and SPO2 sensors.
ECG features extraction like QRS complex, R-R interval and heart rate.
Identify abnormal ECG data using an intelligent data analysis scheme based on a new cosine similarity
measure.
42
4.2.3 The Proposed Similarity Measure for Exploring Abnormal ECG Signal
ECG signal can be interpreted by an experienced cardiologist to diagnose a wide range of possible arrhythmias.
ECG has been used as diagnostic tool and long been a research area of interest in the physiology field, because the
ECG reflects a persons state of health states. Also the ECG is cyclic signal and changes with time, it belongs to
the sequential data type. So we apply a similarity measure scheme considering sequential semantics to measure the
similarity degree between any the recived ECG signal and aset of stored ECG signals for examining abnormal ECG
sequences. To identify abnormal ECG signals, this study employed the cosine similarity measure to compute the
similarity degree between any two ECG sequences. Suppose two ECG sequences P1 and P2 contain totally k samples.
The similarity degree between two ECG sequences measured by the cosine measure can be formulated as:
sim (P1 , P2 ) = q
(1)
Where P1 = (s11 , s12 , ... , s1i , ... , s1k ) and P2 = (s21 , s22 , ... , s2i , ... , s2k ) represent, respectively the sample
vectors of the ECG sequence P1 and the ECG sequence P2 , s1i and s2i stand for the ith sample in the ECG sequence
P1 and the ECG sequence P2 , respectively. However, the cosine similarity measure cannot identify the similarity
44
degree between any two sequences well, because it cannot completely identify a symbolic semantics. For example,
the similarity degree of the sequence ABCD and DCBA measured by cosine measure is equal to 1 since both
the sequences include the same set of symbolic terms. This similarity degree evaluated by cosine measure implies
that both sequences are completely identical in terms of semantics, but they have completely different meanings.
In other words, the ECG sample that appears in different positions represents a different meaning. To consider a
samples position for promoting the accuracy when identifying abnormal ECG sequences, the proposed hamming
distance was applied to modify the cosine measure for measuring the similarity degree between two ECG sequences
with increased precision. The proposed hamming distance for both the ECG sequences can be computed using the
following formula:
m
(2)
i=1
Where L(s1i ) and L(s2i ) represents respectively the location index values of the ith samples that appears simultaneously in both the ECG sequences P1 and P2 , m is the total number of samples that appear simultaneously in both
the ECG sequences P1 and the ECG sequence P2 . To simultaneously consider both similarity degree and sequence
semantics when identifying abnormal ECG sequences, the new cosine measure is expressed as:
nsim (P1, P2 ) =
H (P1, P2)
1
& sim (P1 , P2 )
MAX (H)
(3)
Where Max(H) represents the maximum hamming distance among the ECG sequence P1 with all identifying ECG
sequences. To determine whether the proposed new cosine similarity measure correctly explores abnormal ECG
sequences, this study measured 10 actual ECG signals from 10 different people to compare their similarities to each
other using the original and new cosine similarity measures. The aim of this experiment was to determine whether
the proposed new cosine similarity measure can correctly identify the two most similar ECG patterns from a human
ECG physiological database. This will be helpful for judging whether a measuring ECG is similar to any abnormal
ECG signalsstored in the human ECG physiological database. Figure 9 shows four ECG signals measured from four
different people. Table 2 and Table 3 show the similarities measured using the cosine measure and the proposed new
cosine measure, respectively, for ten ECG signals stored in the server database. The experimental results shown in
Table 2 demonstrate that ECG 1 is the most similar to ECG 3. However, Figure 9 obviously indicates that ECG 1 is
the most similar to the ECG 2, not to ECG 3. The experimental result listed in Table 3 confirms that the proposed
new cosine similarity measure can correctly identify which ECG sequence is the most similar to a comparative ECG
sequence, since the position semantics of ECG sequence are considered.
10
ECG 1
0.9974
0.9987
0.9963
0.9975
0.9976
0.9930
0.9980
0.9970
0.9985
ECG 2
0.9974
0.9963
0.9954
0.9972
0.9961
0.9914
0.9968
0.9957
0.9971
ECG 3
0.9987
0.9963
0.9949
0.9976
0.9961
0.9922
0.9970
0.9957
0.9972
ECG 4
0.9963
0.9954
0.9949
0.9962
0.9950
0.9904
0.9956
0.9943
0.9956
ECG 5
0.9975
0.9972
0.9976
0.9962
0.9974
0.9927
0.9980
0.9972
0.9986
ECG 6
0.9976
0.9961
0.9961
0.9950
0.9974
0.9915
0.9968
0.9960
0.9974
ECG 7
0.9930
0.9914
0.9922
0.9904
0.9927
0.9915
0.9927
0.9912
0.9937
ECG 8
0.9980
0.9968
0.9970
0.9956
0.9980
0.9968
0.9927
0.9966
0.9978
ECG 9
0.9970
0.9957
0.9958
0.9943
0.9972
0.9960
0.9912
0.9966
0.9969
ECG 10
0.9985
0.9971
0.9972
0.9956
0.9986
0.9974
0.9937
0.9978
0.9969
10
ECG 1
0.0161
0.0155
0.0064
0.0053
0.0014
0.0006
0.0007
0.0089
ECG 2
0.0161
0.0086
0.0069
0.0102
0.0022
0.0003
0.0006
0.0003
0.0100
ECG 3
0.0155
0.0086
0.0028
0.0042
0.0011
0.0102
ECG 4
0.0064
0.0069
0.0028
0.0050
0.0041
0.0011
0.0030
0.0011
0.0036
ECG 5
0.0053
0.0102
0.0042
0.0050
0.0044
0.0003
0.0125
ECG 6
0.0014
0.0022
0.0011
0.0041
0.0044
0.0025
0.0039
0.0011
0.0012
ECG 7
0.0006
0.0003
0.0011
0.0003
0.0025
0.0061
0.0085
ECG 8
0.0006
0.0030
0.0039
0.0061
0.0091
0.0006
ECG 9
0.0007
0.0003
0.0011
0.0011
0.0085
0.0091
ECG 10
0.0089
0.010
0.0102
0.0036
0.0125
0.0011
0.0006
signal. The flow chart in Figure 11 explains the operation of PPG processing. Where the developed algorithm first
reads 600 consecutive samples sent by mobile care unit. Since the sampling rate was 5ms, it takes 3 Sec to read the
6000 samples. The DC component minima of 600 samples are subtracted out from the samples. Next, the range of the
samples is computed. If the range is less than 50 counts, the received PPG waveform is very weak, and is considered
to be a noise. This could happen when no PPG signal is detected through fingertip (sensor is faulty or disconnected)
or the gain of the amplifier on Easy Pulse board is set very low. If the range of ADC samples is greater than 50, it is
considered as a valid PPG signal. The samples are scaled to 1-1023 in full swing of display. Next, a 21-point moving
average filter is applied to remove the unnecessary high frequency components usually noise in the PPG signal.
environment by Java language, and is debugged and run on a smartphone with the Android operating system. The
design and implementation of the Android monitoring application include two aspects: design and implementation of
the android application user interface and android application functional modules. Two application user interface
needs to be designed in this android application, one is used for the control and operation of the application, the other
one is used for displaying analog and digital forms of physiological parameters. When the device is connected to the
mobile care unit of thePatient it will automatically update current basic information about the patient.Additionally,
there is possibility to browse offline basic medicaldata, which are stored in the internal database and access to personal,
electronichealthcare records, which contain the list of actual diseasesand prescribed medications, information about
recentevents and a doctors personal notes about the patient as shown in Figure 12(a). Doctor can add new patient or
modify the stored information if needed as shown in Figure 12(b).There is also a possibilityto carry out test directly
from the application.Figure 13 shows an example of real time measurement for patient basic vital signs (ECG, heart
rate, blood pressure, temperature, spo2).
5. CONCLUSION
In this paper, remote patient monitoring system using android mobile phones is presented which allows doctors
to view his patients vital parameter remotely and dynamically in real time. For the patient side, a mobile care unit
is attached to the patients body to acquire and transmit vital signs to the remote server. At server side, MATLAB
application is developed which monitors the received vital signs and apply a set of analysis algorithms to determine
47
(a)
(b)
the patients status. For doctor side, mobile devices with Android OS were chosen. This is one of the most popular
systems (next to the Apple OSs) used in mobile devices, characterized by the following advantages. An open source
platform based on Java, with multitasking and wider hardware support. Personal mobile devices with Android system
are powerful and flexible in use. It reduces both the time and cost needed for system development. For these reasons,
easy to use android application is developed to view the patients vital signs, personal information and medical
48
records. Comparing this system with other systems, it has the following features:
Continuously monitor patients vital health conditions instead of the discrete measurements.
Integrates sensor unit, processing unit and communication unit in one chip bonded to the patients body
called mobile care unit so the patient could do his/her daily activities during monitoring.
At server side, it provides ECG and PPG signal filtering & processing, feature extraction, detection of any
abnormalities in ECG, calculating beat rate and so on using the most familiar and multipurpose MATLAB software.
Android based mobile phones, patient monitoring application is developed which allows doctors to monitor
the health status of the patient. This application also provides alerts, reminders and emergency notifications for vital
measurements to help doctors to take timely decisions in emergency situations.
References
[1] B. Rechel, Y. Doyle, E. Grundy, and M. McKee, How can health systems respond to population ageing?, 2009.
[2] I. Korhonen, J. Parkka, and M. Van Gils, Health monitoring in the home of the future, Engineering in Medicine
and Biology Magazine, IEEE, vol. 22, no. 3, pp. 6673, 2003.
[3] M.-k. Suh, C.-A. Chen, J. Woodbridge, M. K. Tu, J. I. Kim, A. Nahapetian, L. S. Evangelista, and M. Sarrafzadeh,
A remote patient monitoring system for congestive heart failure, Journal of Medical Systems, vol. 35, no. 5,
pp. 11651179, 2011.
[4] A. Hande, T. Polk, W. Walker, and D. Bhatia, Self-powered wireless sensor networks for remote patient
monitoring in hospitals, Sensors, vol. 6, no. 9, pp. 11021117, 2006.
[5] E. Sardini and M. Serpelloni, Instrumented wearable belt for wireless health monitoring, Procedia Engineering,
vol. 5, pp. 580583, 2010.
[6] F. E. Tay, D. G. Guo, L. Xu, M. N. Nyan, and K. L. Yap, MEMSWear-biomonitoring system for remote vital
signs monitoring, Journal of the Franklin Institute, vol. 346, no. 6, pp. 531542, 2009.
[7] H. Mateev, I. Simova, T. Katova, N. Dimitrov, and I. Christov, TEMEOA novel mobile heart rhythm telemonitoring system, in Computing in Cardiology, 2011, pp. 833836, IEEE, 2011.
[8] A. Alesanco and J. Garcia, Clinical assessment of wireless ECG transmission in real-time cardiac telemonitoring, Information Technology in Biomedicine, IEEE Transactions on, vol. 14, no. 5, pp. 11441152, 2010.
[9] W. Walker, T. Polk, A. Hande, and D. Bhatia, Remote blood pressure monitoring using a wireless sensor
network, in 6th Annual IEEE Emerging Information Technology Conference, Citeseer, 2006.
49
[10] C. Rotariu, A. Pasarica, H. Costin, F. Adochiei, and R. Ciobotariu, Telemedicine system for remote blood
pressure and heart rate monitoring, in E-Health and Bioengineering Conference (EHB), 2011, pp. 14, IEEE,
2011.
[11] P. Pandian, K. Mohanavelu, K. Safeer, T. Kotresh, D. Shakunthala, P. Gopal, and V. Padaki, Smart Vest:
Wearable multi-parameter remote physiological monitoring system, Medical Engineering & Physics, vol. 30,
no. 4, pp. 466477, 2008.
[12] C. Rotariu, H. Costin, I. Alexa, G. Andruseac, V. Manta, and B. Mustata, E-Health System for Medical
Telesurveillance of Chronic Patients, International Journal of Computers Communications & Control, vol. 5,
no. 5, pp. 900909, 2010.
[13] V. Gay and P. Leijdekkers, A health monitoring system using smart phones and wearable sensors, International
Journal of ARM, vol. 8, no. 2, pp. 2935, 2007.
[14] J. Rodriguez, A. Goni, and A. Illarramendi, Real-time classification of ECGs on a PDA, Information Technology in Biomedicine, IEEE Transactions on, vol. 9, no. 1, pp. 2334, 2005.
[15] C. Wen, M.-F. Yeh, K.-C. Chang, and R.-G. Lee, Real-time ECG telemonitoring system design with mobile
phone platform, Measurement, vol. 41, no. 4, pp. 463470, 2008.
[16] G. Tartarisco, G. Baldus, D. Corda, R. Raso, A. Arnao, M. Ferro, A. Gaggioli, and G. Pioggia, Personal health
system architecture for stress monitoring and support to clinical decisions, Computer Communications, vol. 35,
no. 11, pp. 12961305, 2012.
[17] C. Chan, C. Chen, W. Chou, Y.-L. Ho, Y.-H. Lin, and H.-P. Ma, Live demonstration: A Mobile ECG Healthcare
Platform, IEEE Biomedical Circuits and Systems Conference, p. 87, 2012.
[18] P. Leijdekkers, V. Gay, and E. Lawrence, Smart homecare system for health tele-monitoring, in Digital Society,
2007. ICDS07. First International Conference on the, pp. 33, IEEE, 2007.
[19] D. Lou, X. Chen, Z. Zhao, Y. Xuan, Z. Xu, H. Jin, X. Guo, and Z. Fang, A Wireless Health Monitoring System
based on Android Operating System, IERI Procedia, vol. 4, pp. 208215, 2013.
[20] X. Guo, X. Duan, H. Gao, A. Huang, and B. Jiao, An ECG Monitoring and Alarming System Based On
Android Smart Phone, Communications and Network, vol. 5, p. 584, 2013.
[21] V. Jones, A. van Halteren, N. Dokovsky, G. Koprinkov, R. Bults, D. Konstantas, R. Herzog, et al., Mobihealth:
Mobile health services based on body area networks. Springer, 2006.
[22] B. Mehta, D. Rengarajan, and A. Prasad, Real time patient tele-monitoring system using LabVIEW, Int. J. Sci.
Eng. Res, vol. 3, pp. 111, 2012.
[23] P. K. Gakare, A. M. Patel, J. R. Vaghela, and R. Awale, Real time feature extraction of ECG signal on android
platform, in Communication, Information & Computing Technology (ICCICT), 2012 International Conference
on, pp. 15, IEEE, 2012.
[24] Z. Aihua and H. Linghao, The system of pulse monitoring based on Windows Mobile, in Business Management
and Electronic Information (BMEI), 2011 International Conference on, vol. 4, pp. 519522, IEEE, 2011.
[25] V. Kumar, A novel approach to pattern recognition in real-time arrhythmia detection, in Engineering in
Medicine and Biology Society, 1988. Proceedings of the Annual International Conference of the IEEE, pp. 78,
IEEE, 1988.
[26] H. Dickhaus, J. Gittinger, and C. Maier, Classification of QRS morphology in Holter monitoring, in [Engineering in Medicine and Biology, 1999. 21st Annual Conference and the 1999 Annual Fall Meetring of
the Biomedical Engineering Society] BMES/EMBS Conference, 1999. Proceedings of the First Joint, vol. 1,
pp. 270vol, IEEE, 1999.
50