mDROID - An Affordable Android based mHealth System

Anju Kansal, Avval Gupta, Kolin Paul and Sanjiva Prasad

Computer Science and Engineering, IIT Delhi, Hauz Khas, 110016, New Delhi, India

Keywords:

mDROID, mHealth, Patient Record System, Biomedical Sensors, Medical Data, XML Medical File, Server,

Bluetooth, Android.

Abstract:

This paper presents a novel approach towards the development of a portable, user-friendly and affordable

mHealth System using Android based mobile devices. The Android application captures personal, biometric

and medical data of a patient and combines them to generate an XML based medical record which is then

uploaded to a server. The medical data are gathered from different sensors interfaced to an Arduino UNO

board. The data are then converted into packets, which are transmitted on request to the mobile phone using

Bluetooth. The XML report makes the mDROID system robust for query and inter-operability at the server end

and also for merging with standardized XML based medical repositories. The mDROID system is designed

for use by minimally trained health workers in the ﬁeld.

1 INTRODUCTION

Providing affordable and scalable healthcare is one

of the most important development goals, and one

which may pose many technological challenges. Re-

search by Cisco (CVNI, 2013) projects that by the

end of the year 2016, there will be approximately

10 billion mobile devices in use around the world.

mHealth refers to using mobile phone technology in

providing healthcare. In recent years, mHealth has

shown promising growth mainly because of the fact

that it can alter how healthcare is delivered, the qual-

ity of patient experience, and the cost of health care

(Wang and Liu, 2009). The most common activity

in mHealth has been the creation of health call cen-

tres, which respond to patient inquiries. This was fol-

lowed by using SMS for appointment reminders, us-

ing telemedicine, accessing patient records, measur-

ing treatment compliance, raising health awareness,

monitoring patients, and physician decision support.

With advances in technology, wearable sensors

have been integrated with mobile phones for mon-

itoring health. Despite mHealth having several ad-

vantages in personal healthcare, a 2011 global survey

of 114 nations undertaken by the World Health Orga-

nization found that mHealth initiatives established in

many countries have several issues resulting in vari-

ation in adoption (WHO, 2011). mHealth has shown

maximum adoption in developed countries mainly be-

cause mHealth applications for personal health are

available only on high-end expensive mobile phones.

Thus, mHealth has grown in developed countries

whereas it is still at an exploratory stage in under-

developed nations (Mechael et al., 2010).

In developing countries, data are manually col-

lected by community health workers for (rural) Health

Centres. These data are collected in the ﬁeld and writ-

ten “in rough” on paper. Later, these readings are

manually uploaded onto servers. The various stages

require constant scrutiny to ensure reliability (Otieno

et al., 2012). Manual entry makes the whole system

both cumbersome and vulnerable to human errors.

Moreover, an important aspect of any mHealth imple-

mentation is ensuring that the data are exchangeable.

Existing systems frequently overlook the fundamental

requirements of interoperable data standards (WHO,

2012).

All the factors discussed above demand an

mHealth solution which (1) has an automated sys-

tem for collection of medical data by health workers

with minimal training, (2) consists of all basic med-

ical sensors and is yet portable, (3) is affordable for

mass usage in under-developed countries, and (4) col-

lects data in a format that can be easily exchangeable

and usable within heterogeneous systems. The pro-

posed mDROID system has been designed to address

all these aspects keeping in mind the “pain-points” of

community workers. The all-in-one mDROID system

uses an Android based device to capture medical data

through various electronic sensors with ease, com-

109

Kansal A., Gupta A., Paul K. and Prasad S..

mDROID - An Affordable Android based mHealth System.

DOI: 10.5220/0004803501090116

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2014), pages 109-116

ISBN: 978-989-758-010-9

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: End-to-End Data Flow of the mDROID System.

bines them with personal and biometric information

of the patient to form an XML based medical report,

which can be uploaded to the server on request.

Technologies like mDROID will facilitate heath-

care workers in monitoring the health of local popu-

lations at regular intervals, especially at the bottom of

the pyramid, thereby empowering healthcare ofﬁcials

to track epidemics and take necessary steps to prevent

them. It will also allow policy makers to understand

the correlation between health and environmental fac-

tors and make changes to suit the population of the

locality.

2 METHODOLOGY

The mDROID system is broadly divided into two ma-

jor parts. The ﬁrst part consists of getting medical

data from various sensors on the Android phone. The

second part involves development of an Android ap-

plication for processing the data received and gener-

ating medical report to be sent to a server. Figure 1

depicts an overview of the data ﬂow from sensors to

the server.

2.1 Obtaining Medical Data from

Sensors to Android Device

The mDROID system supports multiple medical sen-

sors. The system was designed to make the sensors

work efﬁciently in capturing multiple data simulta-

neously rather than a one-by-one approach, which is

time consuming, inefﬁcient and possibly less reliable.

Concurrent capture is achieved by using an interme-

diate platform that is capable of connecting to mul-

tiple sensors simultaneously and transmitting to the

Android based device. The data from multiple sen-

sors are combined to form packets, which are then

transmitted to the Android device through one of the

connectivity options supported by Android and Ar-

duino. The mDROID system uses Bluetooth as the

mode of communication between the two platforms.

The two main components of the microcontroller pro-

gram are: (1) formation of medical data packets, and,

(2) managing data transmission through Bluetooth.

2.1.1 Formation of Medical Data Packets

The clinical data obtained from the sensors are cate-

gorized into (1) discrete data points, and (2) contin-

uous data streams. The former category comprises

the measurement of discrete physiological parameters

such as temperature, pulse rate, oxygen saturation in

blood etc., whereas the latter covers measuring graph-

ical medical parameters such as ECG, which is re-

quired to be plotted on a graph over an uninterrupted

time interval.

While acquiring the discrete parameters, the mi-

crocontroller program fetches different data points

from multiple medical sensors and couples them to-

gether into data packets. This feature allows all the

sensor data points collected at one time to be trans-

mitted together. Coupling readings from different

sensors makes it possible to correlate the different

physical parameters. Each data packet consists of a

frame start delimiter, end delimiter, sensor identiﬁers,

sensor data start and end delimiters, and actual sen-

sor data, as depicted in Figure 2. These delimiters

are required because some bytes are often lost dur-

ing transmission in Bluetooth communication. The

packet Start and End delimiters help in distinguishing

complete data packets from incomplete ones, which is

necessary in building a robust system. Sensor identi-

ﬁers are used to label particular sensors’ data. This al-

lows the mDROID system to form custom data pack-

ets depending on the speciﬁc sensors requested by

the user application, supporting ﬂexibility in the set

of sensors used by an application and eliminating the

need for maintaining a ﬁxed sequence of sensors in

the data packets.

The continuous data stream is acquired separately

without coupling the values with any other sensor

data.

HEALTHINF2014-InternationalConferenceonHealthInformatics

110

Figure 2: Design of Data Packet.

2.1.2 Data Transmission through Bluetooth

The microcontroller program controls what is to be

transmitted and when. The system continuously lis-

tens for a request for a data packet or continuous

stream of data identiﬁed by a REQUEST byte. On re-

ceiving a request byte, it identiﬁes whether the request

is for a discrete data packet or a continuous stream

of data. In the former type case, it fetches the val-

ues from the sensors, couples them into a data packet,

transmits the packet to the application, and goes back

to the listening state. In case of a request for a con-

tinuous stream, it starts fetching and transmitting a

stream of values, until it receives a STOP request af-

ter which it goes back to the listening state.

2.2 Android Application

The second major part of the mDROID system con-

sists of the Android application that forms the core of

the system.

Figure 3: Overview of the mDROID Application.

Figure 3 shows a broad overview of the Android

application. The app handles the Bluetooth connec-

tivity with the Arduino UNO board for fetching the

medical data packets/data stream from it.

Depending on the requirement, the app requests

for a data packet or a stream of data from the Arduino

Figure 4: Medical Data Acquisition in Android through

Bluetooth.

board. In case of data packets, the application de-

composes them to obtain the data of individual sen-

sors. The design of the application allows the data

to be viewed on the application screen individually

or collectively through a pragmatic interface. The al-

gorithm for obtaining the data from a data packet is

depicted in Figure 4.

In case of a stream of data, the application ﬁrst

sends the corresponding request to the Arduino board,

starts receiving the data stream, plots a graph for a

speciﬁc time interval, and sends a stop request to the

board on completion of the time interval.

mDROID-AnAffordableAndroidbasedmHealthSystem

111

In addition to medical data from the sensors, the

application also collects personal information of the

patient, viz., name, age, sex, phone number etc.

The application also utilizes the integrated camera

of the device to capture patient image, which serves as

biometric information of the patient. This feature in-

troduces a sense of authenticity to the system in that it

will make it hard for a health worker to forge medical

records.

Once the personal, biometric and medical data

of the patient are obtained, the application combines

them to generate an XML based medical report. The

application also allows for this report to be sent to a

server on network availability, completing the end-to-

end data ﬂow.

3 IMPLEMENTATION

The components used in the implementation of the

mDROID system were carefully selected keeping in

mind portability and low power consumption. Also,

special emphasis was paid to design the user interface

of the mobile phone application in such a fashion that

a minimally trained health worker can capture and up-

load data with ease.

3.1 Medical Sensors and Arduino

Program

In the current version of the mDROID system, the

main parameters captured are Body Temperature,

Pulse Rate, Oxygen Saturation, Galvanic Skin Re-

sponse and ECG.

3.1.1 Body Temperature Sensor

The body temperature was captured using a sensor

which has a TMP102 digital temperature sensor at its

kernel. It gives a 12-bit, 0.0625

◦

C resolution and pro-

vides an accuracy of 0.5

◦

C. It can monitor tempera-

ture ranging between -25

◦

C to +85

◦

C. Since this sen-

sor works at a voltage range of 1.4V to 3.6V DC and

provides an easy two-wire serial interface, it works as

a perfect match for mDROID.

3.1.2 Oxygen Saturation and Pulse Oximeter

Pulse oximetry is a non-invasive method for indicat-

ing the arterial oxygen saturation. Oxygen satura-

tion is deﬁned as the amount of oxygen dissolved in

blood. To measure oxygen saturation, two different

light sources with wavelength of 660 nm (red light)

and 940 nm (infra-red light) are used. By measuring

the absorption spectrum of the two light sources, oxy-

gen saturation is measured. The pulse oximeter used

for mDROID system works at 3.3Volts.

Pulse rate measures the average number of pulses

per minute. The pulse rate is obtained from the pulse

oximeter sensor which measures the peak to peak

time interval between the pulses and calculates the av-

erage.

3.1.3 GSR - Galvanic Skin Response

GRS is recorded by a sensitive ohm-meter which

measures the electrical conductivity of the skin.

Sweat glands which are under the control of sympa-

thetic nervous system change the conduction level of

the skin with variation in the psychological state of

the person. The sensor for GSR for the mDROID sys-

tem works at 3.3 Volts which allows portability and

can run easily on batteries.

3.1.4 ECG - Electrocardiogram

ECG is one of the most common tools useful to mon-

itor the status of health of the heart. For a detailed

study, a 12 lead ECG is preferred but arrhythmia can

be monitored from a single lead ECG. mDROID is

currently designed to capture a 10 second single lead

ECG data in the form of an image.

The system uses an INA321 instrumentation am-

pliﬁer at the kernel. A second order ﬁlter was utilized

to ﬁlter the noise and capture the ECG in the monitor-

ing mode (critical band of 0.5Hz - 40 Hz). The data

were sampled at 250 samples per second. These data

are saved in the form of an image in the mDROID app.

3.1.5 Arduino Program

The mDROID system uses the Arduino UNO R3

board. The code is written using the open source Ar-

duino 1.0.4 software. The e-Health Sensor Shield and

e-Health library is used for interfacing the Arduino

board to the medical sensors. RN-42 Bluetooth Mod-

ule is used for the Bluetooth transmission using the

Serial Port Proﬁle (SPP). Any device communicating

through Bluetooth is required to ﬁrst enter the pass-

code for pairing with the hardware. Once a device

has been paired, it can start communicating.

3.2 Android Application

The mDROID application is developed using the An-

droid SDK 2.3.3 Java platform. Android was cho-

sen for the implementation of the system owing to its

widespread use, affordability, and extendability. The

HEALTHINF2014-InternationalConferenceonHealthInformatics

112

application supports any device running Android ver-

sion 3.0 and above with Bluetooth 2.0.

There are four major components of the Android

application: (1) obtaining patient information, (2) ob-

taining patient image, (3) obtaining medical data, and

(4) combining all data into a medical report to be up-

loaded to server.

3.2.1 Obtaining Patient Information (Personal

Data)

The ﬁrst component in the patient record system is ob-

taining the general information about the patient. As

soon as the user launches the app, he/she is displayed

a form to be ﬁlled on the app. The module collects

the name, age, sex, and phone number of the patient

and performs suitable validations on the data entered

before allowing the user to proceed. The various val-

idations imposed by the system are as follows

• All the entries must be ﬁlled. No ﬁeld can be left

blank.

• Name and Sex can take text values. Age and

phone number should be numeric values.

• Sex can take only M or F as values.

• Phone number must be 10 digits.

3.2.2 Capturing Image (Biometric Data)

After properly entering acceptable values in the per-

sonal data of the patient, the user is navigated to cap-

ture an image of the patient, which serves as biometric

information of the patient. Once an image is captured,

the user has the option of either capturing another im-

age (if he/she is not satisﬁed with the current image)

or proceeding to capture the medical data. The latest

captured image is taken into account.

The image name contains the timestamp when

the image was captured and is of the form

“IMG yyyymmdd hhmmss X.jpg”, where X is a ran-

dom number. This allows the image to be uniquely

identiﬁed on the server by minimizing the probability

of different image ﬁles having the same name.

3.2.3 Obtaining Medical Data

This component is responsible for establishing Blue-

tooth communication with the sensor hardware and

fetching medical data from the sensors. After the per-

sonal and biometric data are successfully acquired in

the application, the user is navigated to proceed to col-

lection of medical data. The user is required to select

The speciﬁc choices described below only illustrate

that these data must be checked for conformance to a spec-

iﬁed format

the desired hardware from a list of available Bluetooth

devices to initiate a connection. Once the connection

is established, the user can hit a button to begin the

data exchange with the microcontroller.

The medical data are displayed to the user in the

app. The design of the GUI allows the user to view

all the discrete medical data at once or particular data

individually. Figure 5 shows a screenshot of the An-

droid app. The medical data displayed are the aver-

ages of the latest ﬁve values retrieved from the sen-

sors. Every time a new value is received, the average

is updated and used. The implementation allows cap-

turing of a 10 second ECG of the patient. The ECG

data is plotted separately and is saved as an image.

Figure 5: Screenshot of mDROID App.

3.2.4 Generating XML Record

Medical record generation is another crucial aspect

of the mDROID system. The design of the medical

record is based on XML because of its simplicity of

representation, versatility and widespread acceptance

as a vehicle for information exchange in the modern

era.

This component is used to put all the pieces of the

system together. After the acquisition of desired med-

ical data, the application generates a complete medi-

cal record based on the user session and stores it on

the Android device. A sample medical record gen-

erated by the application is depicted in Figure 6. If

the user wishes, he/she can choose to upload this data

record onto the server.

The xml record along with the captured image ﬁle

and ECG image is uploaded on the server, where the

mDROID-AnAffordableAndroidbasedmHealthSystem

113

Figure 6: A Sample XML Report Generated by the

mDROID App.

images are stored in separate directories. The pa-

tient img and ecg img tags in the xml report contain

the paths of the respective image ﬁles on the server.

4 APPLICATIONS

The portable and user friendly mDROID system is de-

signed to help acquire data of the patients by health-

care workers in the ﬁeld. The mobile phone based

system helps the health workers in acquiring patient

data without any specialized or intensive training.

The mDROID application automatically converts

all the acquired data of the patient into an XML ﬁle,

which is uploaded to the server. XML allows easy

consolidation, transmission and retrieval of medical

records from servers. Moreover, it allows existing

protocols to be used to exchange data among hetero-

geneous systems. The capability of XML to exchange

data among different types of computing platforms

enables disparate back-end systems in healthcare or-

ganizations to work together. The data records gen-

erated by the mDROID system can hence be easily

made available to the care givers and researchers.

5 DISCUSSION

Over the recent years, mHealth is increasingly be-

coming a prime focus of the medical commu-

nity, leading to intense work in this ﬁeld. The

mDROID system presents a simple user interface with

a straightforward workﬂow to capture the various

health parameters of patients. The choice of plat-

form for the implementation was Android, owing to

its various advantages. Android based devices are

widely used and their availability is rapidly increas-

ing over a broad range of cost and features, lead-

ing to smartphones and tablets being available at very

low prices. According to the International Data Cor-

poration (IDC) Worldwide Quarterly Mobile Phone

Tracker, the Android platform crossed 80% of the

smartphone market share in the third quarter of 2013

(IDC, 2013). Moreover, Android allows for the built-

in utilities of the smart devices to be utilized without

having to design a separate control processor. The

mDROID system is focused on integrating multiple

medical devices onto a single platform making an all-

in-one package for medical data collection. This re-

duces the cost effectively when compared with using

standalone devices for measuring each of the medical

parameters. There are a host of independent medi-

cal peripherals in the market for the iPhone (iPhone

Medical Devices, 2013). However, the cost of an

iPhone as well as these medical add-ons is uneconom-

ical and hence cannot be used as a cost-effective so-

lution for the health workers working at the bottom

of the pyramid. Moreover, all of these individual de-

vices are very speciﬁc and cannot be used together

with other sensors in a customisable fashion. The

mDROID system has been designed to collect data

from medical sensors while keeping them separate

from the phone. The use of Bluetooth enables easier

integration of newer sensors as well as replacement of

sensors. This creates a potential for opening up of a

market for many more such affordable sensors. The

mDROID is a novel one-of-a-kind customisable sys-

tem in mobile healthcare. XML based medical reports

add the beneﬁt of portability of medical data across

platforms. Biometric information brings trust and au-

thenticity to the system.

The Swasthya Slate (Health Tablet) developed by

PHFI (Swasthya, nd) is a similar system which works

on Android based tablet phones. It enables the acqui-

sition of clinical data through electronic sensors. The

price of the device costs approximately INR 30,000.

It has a couple of additional sensors such as test for

water quality, which we plan to deploy in our project

in a later stage. However, there is no provision for

capturing patient image and other metadata, or for

the generation of XML based medical reports in the

Swasthya Slate. The mDROID system is more ﬂexi-

ble as it works on all mobile Phones running the An-

droid platform, including the inexpensive ones; the

price for the mDROID device is projected to be more

affordable than the Swasthya Slate.

Another framework for mobile healthcare is the

SANA technology developed at MIT (SANA, 2010).

SANA is a reliable tool that allows health workers

to transmit medical ﬁles in the form of text, image,

HEALTHINF2014-InternationalConferenceonHealthInformatics

114

audio, video etc. through a cell phone to a cen-

tral server for archiving and incorporation into an

electronic medical record, and to a remote specialist

for real-time decision support. The prime focus of

SANA tool is to transmit the medical information to

the reach of various stakeholders in the healthcare do-

main. However, there is no support in SANA for uti-

lizing electronic sensors within the mHealth system,

and the user is required to feed the data manually into

the Android application making the application prone

to human errors.

Open Data Kit (ODK, 2009) is a framework for

building user interfaces, collecting data on mobile

devices, and aggregating them onto a server. The

ODK framework supports generic interfaces for a

class of sensors, both internal and external to an An-

droid device, although at the time of implementa-

tion of mDROID, these were not released for pro-

duction. While ODK has been used in some health-

related projects, it is not clear to what extent the type

classes of discrete and continuous medical data can be

supported in the systematic manner we have outlined

above.

This comparative analysis suggests that mDROID

promises to be a more effective tool for health work-

ers when collecting data, with authentication, at the

bottom of the pyramid in developing countries. While

the data authenticity is not foolproof, it offers some

level of conﬁdence that the data was genuinely cap-

tured and not forged.

6 FUTURE WORK

The mDROID system described in this paper relies on

external sensors for obtaining clinical data and is tar-

geted to be operated by non-experts. The next steps

involve providing a framework that can attest to the

provenance of the data collected. The current system

captures the biometric data in the form of patient im-

age, which is not being cross-veriﬁed. Future work

involves implementation of sub-systems for acquisi-

tion and veriﬁcation of biometric information of the

patient as well as the health worker. Further analy-

sis of the system reveals that the clinical data is sent

over a Bluetooth connection to the Android device

and then over a network to the server, both of which

can act as points of network vulnerabilities. This re-

quires investigation of ways to safeguard the privacy

of the medical data over any transmissions. Addi-

tionally, the current system supports a limited range

of medical sensors which are being extended to sup-

port more varieties of sensors, thereby providing a

complete low cost and efﬁcient mHealth data acquisi-

tion system. Further, the current system works using

Bluetooth connectivity. Future scope involves sup-

porting more connectivity modes, viz., Wi-Fi, Zig-

Bee, USB, etc. Currently, the uploaded records are

sent to a small server designed for storage purposes

in XML format. The future work involves support-

ing the server with openEHR standards, and adding

functionalities for extraction and exchange of medi-

cal data from the server. Also, cloud-based solutions

can be used for storing and analyzing medical data

collected.

ACKNOWLEDGEMENTS

This work has been partially supported from the

DeitY (MCIT, Government of India) funded project

RP02597 “Foundations of Trusted and Scalable ‘Last

Mile’ Healthcare”. The fourth author also acknowl-

edges support from a Microsoft Research grant for

precursors to this work. The authors would like to

thank Mr Vijay Kumar for assistance with conﬁgur-

ing the hardware. Dr Shelly Dutta of Sahyog pro-

vided valuable information about the work routines

of healthcare workers in rural Rajasthan.

REFERENCES

CVNI (2013). Cisco visual networking index: Global mo-

bile data trafﬁc forecast update, 2012-2017.

IDC (2013). Idc worldwide mobile phone

tracker, august 7, 2013 press release. http://

www.idc.com/getdoc.jsp?containerId=prUS24257413.

iPhone Medical Devices (2013). 5 Medical de-

vices with the iphone. http://internetmedicine.

com/2013/02/13/medical-device-iphone-smartphon.

Mechael, P., Batavia, H., Kaonga N. andSearle, S., Kwan,

A., Goldberger, A., Fu, L., and Ossman, J. (2010).

Barriers and gaps affecting mhealth in low and middle

income countries. A Policy White Paper commisioned

by The mHealth Alliance.

ODK (2009). The open data kit. http://opendatakit.org/.

Otieno, C., Kaseje, D., Ochieng, B., and Githae, M. (2012).

Reliability of community health worker collected data

for planning and policy in a peri-urban area of kisumu,

kenya. Journal of Community Health, 37:48–53.

SANA (2010). The sana project.

http://sana.mit.edu/wiki/index.php?title=Overview.

Swasthya (n.d.). Swasthya slate of public health foundation

of india. http://www.swasthyaslate.org.

Wang, H. and Liu, J. (2009). Mobile Phone Based Health

Care Technology. Recent Patents on Biomedical En-

gineering, 2:15–21.

mDROID-AnAffordableAndroidbasedmHealthSystem

115

WHO (2011). mhealth: New horizons for health through

mobile technologies. Global Observatory for eHealth

Series, 3.

WHO (2012). Management of patient information: Trends

and challenges in member states. Global Observatory

for eHealth Series, 6.

HEALTHINF2014-InternationalConferenceonHealthInformatics

116