Telemedicine App Development Cost: Budgeting Your Virtual Healthcare Solution
healthphone-Healthcom 2009
1. Mobile Solutions for Front-Line Health Workers
in Developing Countries
Jim Black1
, Fernando Koch2
, Liz Sonenberg3
,
Rens Scheepers4
, Ahsan Khandoker5
, Edgar Charry6
, Brian Walker7
, Nay Lin Soe8
1,8
Nossal Institute for Global Health,
2,3,4
Department of Information Systems,
5,6,7
Department of Electrical and Electronic Engineering,
The University of Melbourne
Melbourne, Australia
{1
jim.black,2
fkoch,3
lizs,4
r.scheepers,5
ahsank,
6
echarry, 7
b.walker, 8
nlsoe}@unimelb.edu.au
Abstract— We introduce an architecture for low-cost mobile
Health (mHealth) applications that run on health-workers’
existing devices. Moreover, we envision extending the phone’s
capabilities with an external to attach “sensor” modules, such as
pulse oximeter, ECG and phonocardiogram. Our design
principles are frugality and simplicity. We propose a
comprehensive solution to aid health-workers in their daily tasks,
at a low-cost and high penetration rate.
Keywords— Mobile computing, mobile health, telemedicine,
sensors, support health-workers.
I. INTRODUCTION
Mobile phones are bridging the digital divide and
transforming many economic, social, and medical realities,
particularly in developing countries. With the penetration of
low-cost handsets and the omnipresence of mobile phone
networks, tens of millions of people who never had a
computer now use mobile devices. On the other hand, trained
health workers and diagnostic testing facilities are a scarce
resource in poor countries’ rural areas, especially in Africa [1].
Despite the billions of dollars invested in health in Africa, the
shortage of appropriate health workers particularly in rural
areas in many countries is a major barrier to health service
coverage for the poor [2].
The growing ubiquity of mobile services allows the
creation of a new generation of electronic health systems
based on mobile computing. Mobile Health (mHealth) is
emerging as an important segment of the field of electronic
health (eHealth) [3] that advocates the utilisation of mobile
technology supporting the next generation health systems. We
suggest that even the simplest solutions would provide a
major contribution to health development in these
communities. It is possible to create a range of mobile phone
applications and low-cost diagnostic devices that will run on
health workers’ own mobile phones, making them useful for
daily activities.
The environment imposes severe restrictions, however.
First, resource constraints mean that we must avoid
introducing new costs, whether capital or recurrent --
developing country health services have tiny per capita annual
budgets! Second, limitations in specialised workforce
availability mean that we cannot rely on distant experts to
interpret data or provide the diagnosis. We expect to operate
in areas with poor radio coverage and the applications will
execute in low-end mobile devices. Finally, due to logistic
limitations we must avoid the need for training to operate the
solution. Therefore, we argue that a model based on data
transmission for distant analysis is unlikely to work.
We aim at non-invasive techniques to measure and report
vital signs and other diagnostic and patient management
information combined with the omnipresence of mobile
phones and the health-workers’ familiarity with these devices.
We are developing ―local applications‖ that run on health-
workers mobile devices. For example: respiratory rate or pulse
rate counter, gestational dates calculator, drug dose calculator,
drip rate calculator, and drug reminder alarm.
Moreover, we are extending the phone’s capabilities with
an external to attach ―sensor‖ modules, such as a pulse
oximeter, ECG and phonocardiogram. These applications
provide a comprehensive solution to common diagnostic and
patient management tasks when combined in the same device.
This work is organized as follows: In the next section we
present our motivation and related work. Section 3 introduces
our proposal. Section 4 presents proof-of-concept
implementations. The paper concludes in Section 5.
II. MOTIVATION AND RELATED WORK
One challenge to provide better healthcare services is how
to increase the participation of front-line health-workers using
limited financial and human resources in order to deliver
assistance to an increasing number of people [4]. Mobile and
wireless technologies can be effectively utilised by matching
infrastructure capabilities to healthcare needs. Research in
mHealth scrutinizes how to take advantage of the wide
availability of mobile devices – e.g. mobile phones, PDA,
2. mobile computers – to deliver mobile health solutions for
front-line workers.
The work in [5] shows that governments, companies, and
non-profit groups are already developing mHealth
applications to improve healthcare. This report presents 51
programs, either currently operating or slated for
implementation in the near future, that are taking place in 26
different developing countries. These applications are creating
new pathways for sharing health-related information, even in
the most remote and resource-poor environments. These
projects provide key solutions in the following areas: (i)
education and awareness; (ii) remote data collection; (iii)
remote monitoring; (iv) communication and training for
healthcare workers; (v) disease and epidemic outbreak
tracking, and (vi) diagnostic and treatment support. Table 1
presents some of these initiatives, as described in [5] and [6].
TABLE I - EXAMPLES OF MHEALTH INITIATIVES
Type Name Description
Analysis,
Diagnosis &
Consultation
Tele-Doc Provides mobile phone devices to
front-line workers, allowing them
(voice) communication with remote
doctors
Analysis,
Diagnosis &
Consultation
Nacer Allows health professionals to
exchange critical health information
with peers; reported data is recorded
in a central database and is available
in real-time for decision-making
Data, Health
Record
Access
AED Satellife Provides support for HIV/AIDS,
malaria, child and maternal health,
and health systems management
programs
Data, Health
Record
Access
EpiHandy Provides a set of tools for collection
and handling of data using mobile
devices
Data, Health
Record
Access
EpiSurveyor Free, open-source software suite to
collect data using handheld
computers and mobile phones
Monitoring /
Medication
Compliance
Cell-Preven Provides real-time data distribution
of symptoms experienced by clinical
trial participants via SMS messages
It is our experience that even people on very low incomes
in developing countries are acquiring and using mobile phones.
They are used for many purposes, including checking market
prices and keeping in touch with relatives who migrate to
urban areas - rarely for health care purposes. Nonetheless, all
but the simplest mobile phones now have operating systems,
and some have very sophisticated and powerful processors.
Therefore, it is possible to explore this capability and write
simple applications that run on front-line health-workers’ own
mobile phones, providing simple ―tools‖ to aid in their daily
activities.
Our aim is to implement low-cost, high penetration
―analysis, diagnosis and consultation‖ solutions that explore
mobile phones’ processing and interfacing capabilities. These
solutions will deliver simple tools to aid front-line workers in
their daily activities.
We have two design principles:
1. frugality, i.e. wherever possible we should avoid
creating new capital or recurrent costs; this means
making minimum or no use of the network
capabilities by concentrating on ―local
applications‖ and low-cost solutions. Moreover,
these applications must run on front-line workers’
own mobile devices, and;
2. simplicity, i.e. the applications and devices must
be as simple as possible; they should require
minimum or no training to operate.
The applications must therefore look and feel as much as
possible like the normal functions of a mobile phone, and
require no more skill than looking up a missed call or adding a
new contact. Moreover, the applications must operate using
minimum computing resources, considering that the available
devices are mostly low-end, inexpensive models.
In what follows we describe the solutions that we are
proposing based on these guidelines.
III. PROPOSAL
Phone Hardware USB
PIM
ECG
Interface
Oximeter
Interface
Pregnancy
Calculator
OtherApps
Oximeter Sensor
(v) A/D Interfaces
ECG Sensor
Phonocardiogram
Microcontroller
Sensors ModuleMobile Phone Module
(vi) USB
Interface
C#
Other Sensor
Phonocardiogram
Interface
RespiratoryRate
Counter
DripRate
Counter
DrugReminder
Alarm
Java
(i)
(ii)
(iii)
(iv)
Fig. 1 - Proposed Architecture
We adopted the following implementation guidelines in
order to achieve our ideals:
Compatibility. The applications and hardware
extensions must be simple and compatible with the
resources available in the mobile phone’s operational
environment such as Operating System features,
programming languages and libraries, and software
and hardware interfaces. This feature facilitates
dissemination to heterogeneous devices and also helps
reduce development and distribution costs.
Integration. The set of applications and hardware
extensions must be integrated in one comprehensive
solution. This feature also facilitates dissemination and
allows, in future work, to integrate the diverse data
sources in more complex applications such as
intelligent data analysis.
Standardization. Both applications and hardware
extensions must comply with existing standards such
as programming languages and libraries and hardware
interfaces. This feature helps to curb development
costs and supports dissemination.
3. Minimalism. The applications must be as simple as
possible, and should require minimum or no training to
operate. In addition, they must operate using minimum
computing resources, considering that the available
devices are low-end, inexpensive models.
Fig. 1 depicts our proposed architecture based on these
principles. It is composed of two modules:
1. Mobile Phone Module, which is provided by the
device’s hardware and software structure; this is
composed of (i) the programming language available
for the device – e.g. Java or C#; (ii) the existing
application and interface methods, for example the
Personal Information Management (PIM) system,
which provides calendar, messaging interface,
notification and other functionalities, and interfaces to
attached hardware; (iii) the external interface, e.g. the
USB port; this is where the developed applications are
being plugged in.
2. External Sensor Module, which provides the interface
to attach external sensors to the mobile phone; it is
composed of (iv) a microprocessor that controls (v) a
number of external Analogue/Digital (A/D) ports
where the sensors are attached, and; (vi) the USB
interface to connect to the mobile phone; once
plugged into the phone, it is recognised as a standard
USB devices.
The composition is intuitive. We develop applications
using the programming language available in the mobile
phone. Depending on the application, it might interface to
other elements of the device, such as the PIM system to issue
notifications (e.g. drug reminder). Other applications interface
to the device’s other peripherals, such as the display, keyboard,
speaker, and camera.
In addition, the A/D ports allow the connection of ―sensor
modules‖ such as the pulse oximeter, ECG, and
phonocardiogram. Other sensors can be developed and
attached to this interface, as long as they respect the electronic
parameters established by the module’s configuration
(described in the next sub-section). The microcontroller is
used to collect data from these ports, pre-process it locally and
forward the information to the USB port. On the mobile phone
device, ―interface applications‖ receive this information and
compose the graphical display.
Next, we introduce the proof-of-concept implementation
and technical details.
IV.PROTOTYPE AND RESULTS
Fig. 2 has a picture of our current prototype. It is composed
of the following elements:
(i) External Sensor Module, in its prototype version: we
intend to produce this module in a reduced form factor
to facilitate distribution and portability; the prototype
version contains the microcontroller (square in the
bottom-left corner), the module’s electronics (square
circuit board on the right), and a ―debug board‖
(underneath the microcontroller) that will not be part
of the final product.
(ii) USB Interface: provided as output by the External
Sensor Module; it requires a common mini-USB cable
to connect this module to the mobile phone.
(iii) Mobile Phone: we use a HTC SmartPhone running
Microsoft Windows Mobile 6.1 for the prototype; we
explain the reasons below.
(iv) Oximeter Probe: attached to one of the A/D ports of
the External Sensor Module, which controls its
functionality such as activating the LEDs and
collecting the results from the light sensors.
We describe the existing applications and external sensor in
the next sub-sections.
Fig. 2 - Prototype
A. Mobile Phone Applications
Mobile phone applications use the programming language
available for the device integrated to the device’s interfacing
capabilities. So far all have been created using C# and the
Microsoft .NET framework, running on SmartPhones using
Microsoft Windows Mobile 6.0. We opted for this
development platform due to hardware availability, Microsoft
Research sponsorship for this project, and because it is
simpler to prototype in this environment. However, we realise
that the health-workers in poor rural areas usually have low-
end, inexpensive mobile phones. Hence, we intend to also
support Java Micro Edition programming language and other
operating systems such as Nokia OS and Symbian OS, and;
(one day) iPhone OS, Android, and others.
We are developing the following applications for the
mobile phone, depicted in Fig. 3.
1) Respiratory and Pulse Rate Calculator: this application
uses the system clock inside the mobile phone to capture
the time that the health worker begins and ends counting
just 10 respiratory cycles, and uses that to calculate the
number of breaths per minute; it provides an accurate
tool to aid health field-workers counting respiratory or
pulse rates.
(i)
(ii)
(iii)
(iv)
4. Fig. 3 - Mobile Phone Applications Prototypes
2) Gestational Dates Calculator: this application invites
the midwife to record the calendar date of the onset of
the pregnant woman's last normal menstrual period. It
then calculates the gestational age today, and the
estimated date of delivery. The application gives the
midwife the option of entering the current gestational
age in terms of lunar cycles. From this it converts to the
solar calendar and estimates the date of the last period,
and the probable date of delivery.
3) Formulary/Drug Dose Calculator: This application
records a subset of the information in the local formulary
- the names, indications for use, dosing regimens and
presentations of drugs available for health workers to
prescribe to their patients. When the health worker
selects a drug, an indication and a presentation (capsules,
tablets, ampoules, etc), the application calculates the
appropriate dose for that patient. The application reports
the therapeutic aim (e.g. "20 to 40 mg/kg/day divided
into four doses") and then tells the health worker exactly
what to write on the prescription pad (e.g. "Amoxicillin
tablets 500 mg, 2 four times per day for seven days").
4) Drip Rate Calculator: This application prompts the
user for the volume to be infused and the infusion period.
It then calculates the corresponding number of drops per
minute. Then the screen begins to flash at exactly that
rate. By holding the mobile phone alongside the giving
set, the health worker need only adjust the flow until one
drop falls every time the screen flashes. No need for
calculations or a wristwatch.
5) Drug Reminder Alarm: This application makes use of
the built-in digital camera that is present in mid-range
and more sophisticated phones. It is meant to be used by
pharmacists when they dispense complex drug regimens.
The pharmacist lays out the correct number of tablets or
capsules to be taken at a given time of day, takes a
digital photograph, and then adds the photo and the time
details to the application's task list. As many different
photos can be added, for as many different times each
day, as necessary. The patient uses the phone as usual,
but when one of the pre-set times arrives an alert appears
on the screen. The patient can elect to "snooze",
delaying the alert for 5, 10 or 15 minutes, or can select
"Show alert". The application then displays the
appropriate photo and the patient lays out the right
tablets, capsules etc ready to take. The response
selection of the patient is recorded in the log of the
application for the pharmacist to later review.
B. External Sensor Modules
External sensors aim to extend mobile phone’s capability
to support health applications. The design allows the
integration of up to 12 sensors. Virtually any type of sensor
can be plugged as long as it provides a signal output
between 0V and 5V. The electronics are composed of
inexpensive elements that cost less than AUD$10.00 (ten
Australian dollars) in the retail market.
We use the 32-bit MCF51JM128 microcontroller
(≈AUD$4.00), from Freescale Semiconductor Inc. This
component can be programmed using C language. It also
provides integrated flash and RAM memory for the
application and data storage. In addition, it supports
interface to external devices via USB interface.
We are developing the following external sensors.
6) Nossal Oximeter: Fig. 4 presents the application
interface for the Nossal Oximeter; the electronics were
depicted in Figure 2 (iv); oximeters work by
measuring the difference in absorption in two
wavelengths of light; with each pulse of arterial blood
into the fingertip or ear lobe it is possible to calculate
the percentage saturation of haemoglobin with oxygen;
despite their potential usefulness, oximeters are
expensive (e.g. around AUD$500) and rarely seen in
developing countries; the primary application for this
product is likely to be in the diagnosis and assessment
of severity of respiratory disease (especially
pneumonia) in an outpatient setting.
(4) Drip Rate Calculator
(3) Formulary/Drug dose
Calculator
(1) Respiratory and Pulse
Rate Calculator
(2) Gestational Dates
Calculator
(5) Drug Reminder Alarm
5. (6) Nossal Oximeter
Fig. 4- Nossal Oximeter Interface
7) Low-cost ECG: provides the sensor’s electronics and
interface application to measure and display the
electrical activity of the heart; this device will allow a
health worker to display an ECG trace from a patient
on the screen of the mobile phone, with a choice of
leads I, II, and III; a brief period of the ECG trace will
be stored by the phone so that the health worker can
back it up and view interesting complexes again at
leisure.
8) Low-cost Phonocardiogram: provides the sensor’s
electronics and interface application to attach a small
microphone to the patient's chest to record the sounds
of the heart and display on the screen, together with
the pulse rate; as the intensity of the heart sounds
varies with respiration, it is also possible to
automatically calculate and display the respiratory rate.
These last two projects are at the conceptual stage. The
first six projects are prototype ready. Video demonstrations
are available at the project’s web-site (see below).
V. CONCLUSION
The main driver for this project is to provide simple,
useful tools to health-workers in remote and underserved
areas. Despite their extreme working conditions, they do a
great job - compensating for a lack of resources with plain
hard work. If we can help to improve their work conditions
and effectiveness in even a small way with the tools
envisioned here, then we have achieved our goal.
Mobile phone technology can provide the basis for a new
generation of affordable, easy to distribute electronic health
solutions for resource-poor communities. We argued in
favour of ―local applications‖ as we do not believe that the
model based on remote analysis would work in this context.
In this project, we extended mobile phone’s processing
and interface capabilities with external sensors to create
low-cost health devices. We are working on five prototype
applications and three external sensors, highlighting the
sub-A$20 Nossal Oximeter device that is currently being
tested.
The proposed architecture allows the plug-in of new
sensors, extending the project’s reach while not adding
substantially to the overall cost.
Our prototype applications have been developed on a
specific platform – the SmartPhone running Windows
Mobile. This was due to availability, ease of use, and
sponsorship. However, we acknowledge that in order to
reach out the intended audience, we need to provide ports to
lower-end platforms more common in underserved
communities. Thus in future we will create a version of our
applications in Java 2 Micro Edition (J2ME). Considering
that the basic structure is the same (displaying techniques,
external hardware interface, etc) and that similar resources
are available between C# and J2ME, we believe that this
objective is achievable with an acceptable level of effort.
Further, before these tools can be widely promoted we
will need to evaluate their acceptability to health workers
and impact on clinical outcomes. Suitable studies are
currently being prepared for African settings.
Finally, the current applications are localised in English
and Portuguese, aiming to be first deployed and evaluated
in Mozambique and Uganda. However, to ensure broader
distribution to other countries, we will produce customised
applications that can be localised to other languages.
This project has been sponsored by kind support from
Microsoft Research. Microsoft does not claim ownership in
the products of this research, but made it a requirement that
any products be made generally available. Thus, we will be
publishing most of the applications and software source
code on the project’s web-site at:
http://www.ni.unimelb.edu.au/ResearchandActivities/Projec
ts/CellPhoneApplications.html
REFERENCES
[1] World Health Organization. 2006. The global shortage of health
workers and its impact. April 2006. Fact sheet 302.
[2] The World Bank. 2008. Health Workers Needed: Poor Left Without
Care in Africa’s Rural Areas. 26 February 2008.
[3] Istepanian, R., Laxminarayan, S. and Pattichis, C. 2006. M-Health. s.l. :
Springer , 2006. p. 623. 978-0-387-26558-2.
[4] Varshney, U. 2007. Pervasive Healthcare and Wireless Health
Monitoring. 2007, Vol. 12, pp. 113-127.
[5] UN Foundation. 2008. mHealth for Development. s.l. : Vodafone
Foundation, 2008.
[6] United Nations. 2007. Compendium of ICT Applications on Electronic
Government -- Mobile Applications on Health and Learning.
Department of Economic and Social Affairs. s.l. : United Nations, 2007.