NOVEL SENSOR TECHNOLOGY INTEGRATION

FOR OUTCOME-BASED RISK ANALYSIS IN DIABETES

Mahesh Subramanian

, Edward C. Conley

1,2

Omer F. Rana

, Alex Hardisty

, Ali Shaikh Ali

Stephen Luzio

, David R. Owens

The Welsh e-Science Centre, Cardiff University School of Computer Science

5 The Parade, Roath, Cardiff, UK, CF24 3AA and

Diabetes Research Unit, Cardiff University School of Medicine

Academic Centre, Llandough Hospital, Penlan Road, Penarth, Cardiff CF64 2XX, UK

Steve Wright, Tim Donovan, Bharat Bedi, Dave Conway-Jones, David Vyvyan, Gillian Arnold

IBM United Kingdom Ltd, Hursley House, Hursley Park, Winchester, UK, SO21 2JN

Chris Creasey, Adrian Horgan, Tristram Cox

Smart Holograms, 291 Cambridge Science Park, Milton Road, Milton, Cambridge, UK, CB4 0WF

Rhys Waite

Zarlink Semiconductor Ltd, Phase 2 Mitel Business Park, Caldicot, Monmouthshire,UK, CF26 5YW

Keywords: Health informatics, home healthcare, biomedical sensor devices, mobility, wearable sensors, decision

support system, individualised risk analysis.

Abstract: Novel sensor-based continuous biomedical monitoring technologies have a major role in chronic disease

management for early detection and prevention of known adverse trends. In the future, a diversity of

physiological, biochemical and mechanical sensing principles will be available through sensor device

‘ecosystems’. In anticipation of these sensor-based ecosystems, we have developed Healthcare@Home

(HH) - a research-phase generic intervention-outcome monitoring framework. HH incorporates a closed-

loop intervention effect analysis engine to evaluate the relevance of measured (sensor) input variables to

system-defined outcomes. HH offers real-world sensor type validation by evaluating the degree to which

sensor-derived variables are relevant to the predicted outcome. This ‘index of relevance’ is essential where

clinical decision support applications depend on sensor inputs. HH can help determine system-integrated

cost-utility ratios of bespoke sensor families within defined applications – taking into account critical

factors like device robustness / reliability / reproducibility, mobility / interoperability, authentication /

security and scalability / usability. Through examples of hardware / software technologies incorporated in

the HH end-to-end monitoring system, this paper discusses aspects of novel sensor technology integration

for outcome-based risk analysis in diabetes.

1 INTRODUCTION

Continuous sensor-based monitoring technologies

are central to new models of ‘proactive’ health and

social care. In healthcare, ‘proactive’ implies a shift

away from ‘reactive’ care – i.e. an ‘illness-centric’

model where interventions are made following

presentation of symptoms or complications. The

‘proactive’ model embraces the World Health

Organisation’s (WHO) definition of health as “a

state of complete physical, mental and social well-

being and not merely the absence of disease or

infirmity” (WHO, 2007). To move towards this

visionary goal, individuals need realistic

119

Subramanian M., C. Conley E., F. Rana O., Hardisty A., Shaikh A., Luzio S., R. Owens D., Wright S., Donovan T., Bedi B., Conway-Jones D., Vyvyan D.,

Arnold G., Creasey C., Horgan A., Cox T. and Waite R. (2008).

NOVEL SENSOR TECHNOLOGY INTEGRATION FOR OUTCOME-BASED RISK ANALYSIS IN DIABETES.

In Proceedings of the First International Conference on Health Informatics, pages 119-126

 SciTePress

opportunities to improve and sustain their health and

quality of life thus contributing to their own well-

being. The availability of validated, relevant and

ubiquitous personal healthcare information to

minimise risk of predictable adverse events can

empower and incentivise individuals to adopt more

healthy lifestyles. Such technology can assist care

throughout the ‘patient path’ (Abidi, 2001;

Shnayder, 2005). Arising from these technology

developments are significant ethical issues - e.g. in

personal data protection and in establishing ethical

authority for personal data reuse. We will discuss

these issues in detail elsewhere in the context of our

Healthcare@Home (HH) project - a research-phase

generic intervention-outcome monitoring framework

that integrates sensor-based technology as part of a

disease early detection and prevention framework.

It is widely anticipated that future health

information systems (HIS) will need to move from

“institution-based” models to those that rely on ‘near

real time’ data integration close to the patient.

Interventions that use ethical risk stratification as

part of a personal data integration framework is a

priority in diabetes, where the number of affected

individuals is predicted to rise from c.135 million

people in 1995 to c.300 million in 2025 (King,

1998). All people with unmanaged diabetes are at

substantially increased risk of serious medical

complications such as retinopathy, kidney failure

and peripheral neuropathy requiring limb

amputation. As part of an individual’s personal

information management, the HH closed-loop model

uses sensor-based trends to compute and stratify risk

in a time frame and operational workflow that is

meaningful and in a format that can be utilised for

building decision support services (DSS). The DSS

model in HH is founded on requirements of the

Diabetes National Service Framework (NSF)

standards for Wales and associated integrated care

pathways (ICP). Section 2 summarises relevant

related work. Section 3 describes technical aspects

of the end-to-end HH system covering (1) smart

sensors; (2) biometric authentication; (3) ‘home hub’

and (4) server-side architecture. Section 4 discusses

concerns arising out of the project and possible

future work with a conclusion in Section 5.

2 RELATED WORK

A healthcare technology platform utilising sensor

devices can underpin comprehensive monitoring

services outside of the hospital environment. This

could support new ways of working that: (1) places

less reliance on frequent clinical visits – subject to

quality control / calibration safeguards and adequate

clinical ‘baseline’ data (2) can incentivise patients to

‘look after themselves’ with realistic (achievable)

personal guidelines within manageable episodes of

care; (3) allows team-based caseload sharing

between clinical visits to monitor progress and make

escalation procedures robust (4) provides for the

development of consistent risk prediction /

longitudinal ‘outcome recording’ methodologies that

are fit-for-purpose in scaleable evidence-based

models (Williams, 2003; Conley, 2007). Several

research projects address the issue of integrated care

through the use of ubiquitous computing devices.

SAPHIRE (Hein 2006) is concerned with

developing a healthcare monitoring and decision

support system for cardiovascular disorders, assisted

by wireless sensor devices in home settings.

(Clemensen 2004) applied pervasive computing

devices to the treatment / monitoring of diabetic foot

ulcers. The CODEBLUE project (Lorincz, 2004) is

typical of sensor device applications in medical

emergency scenarios. In this project, micro-scale

sensor devices (motes) (Crossbow, 2007) have been

used to continuously monitor and wirelessly transmit

vital sign data (e.g. heart rate and oxygen saturation

data) to a data hub for processing. The CART

(continuous automated real-time triage) system,

developed by Advanced Health and Disaster Aid

Network (AID-N) builds on the work carried out by

the CODEBLUE team. A wearable tag has been

developed which performs the following functions:

triage, status display, vital signs monitoring, location

tracking, information display and alarm signalling

(Gao, 2006). The Smart and Aware Pervasive

Healthcare Environment (SAPHE, 2007) supports

telecare and lifestyle monitoring paradigms for early

detection and prevention of adverse events i.e. for

intervention before they become critical or life-

threatening. Several remote healthcare monitoring

systems are that currently use proprietary device

information systems (e.g. Honeywell, American

Telecare and AMD Telemedicine). Many more

device families are expected to be developed in

coming years conforming to global standards being

established by the Continua Healthcare Alliance

companies (Continua, 2007).

The HH system’s conceptual driver is enabling

‘near real time’ risk analysis for early detection and

prevention of disease. A Web Services-based

platform to ‘push’ or ‘pull’ individual’s health-

related data along the patient path is being

HEALTHINF 2008 - International Conference on Health Informatics

120

configured in a manner that will reduce

transcriptional errors. The end-to-end framework

(Figure 1, see below) employs a collection of

clinical hubs, mobile devices and / or dedicated

home-based network servers to one or more data

analysis engines.

3 END-TO-END FRAMEWORK

Figure 1 illustrates the conceptual design of an ‘end-

to-end’ framework adopted in the HH project. The

framework allows data capture from both wearable

sensors and specialist hand-held instruments with

wireless data transfer capability. Through a

messaging fabric and / or dedicated integration

application, raw data used in the risk analysis

modules may originate from a wide variety of

sources and device types (e.g. electronic forms,

physiological monitors, retinopathy cameras,

scanners, clinical chemistry or nucleic acid

sequencing instrumentation). The data is integrated

using a schema indicated as “QUID” (QUantitative

Individualised Data integration) in the figure. In the

context of an ‘initial assessment’ workflow, the

types of diagnostic test to be performed are specified

by a clinical registration procedure to be described

elsewhere. The registration procedure has specific

functions to ensure compliance to ethical protocols

and informed consent procedures for re-use of data

while verifying patient identity (see below). HH has

adopted the Diabetes Continuing Care Reference

(DCCR) dataset as the basis of its schema).

Subsequent risk analysis of baseline data is

undertaken in the context of a “disease model” - a

quantitative research-based predictive framework

that indicates which risk variables are most relevant

to system-specified outcomes. In this regard, HH has

been conceived as a comprehensive healthcare

outcomes evidence-based learning platform.

3.1 Smart Sensors

In monitoring applications, the most common

physiological analytes are blood gases (e.g. carbon

dioxide, oxygen), blood electrolytes (e.g. potassium,

sodium, and chloride), blood glucose, creatinine,

urea, pH, cholesterol, bilirubin and proteins (e.g.

albumin). The relative significance of values

depends on the type of investigative scenario and

progression status of disease. In diabetes, the

measurement of glucose concentration at an

appropriate frequency is routine in short-term

prevention of hyper- and hypo-glycaemic events and

in long-term prevention of complications.

Automated measurements can be enabled to support

continuous monitoring. Conventionally, glucose

monitoring is performed using ‘finger-stick’ devices.

These provide only intermittent measurements and

rely on patient compliance. Next-generation glucose

Figure 1: HH end-to-end framework.

NOVEL SENSOR TECHNOLOGY INTEGRATION FOR OUTCOME-BASED RISK ANALYSIS IN DIABETES

121

monitoring devices aim to provide real-time

continuous measurements by non- or minimally-

invasive means. One such sensor designed for

integration into the HH system uses glucose-

responsive holograms based on thin-film polymers

incorporating phenylboronic acid receptors.

Selective binding of glucose to the receptors induces

swelling or contraction of the film. This physical

change in turn causes the spacing between

holographic fringes within the film to increase or

decrease, thereby modulating the colour of the light

diffracted according to Bragg’s Law.

To obtain repeatable quantitative measurements

(figure 2), the colour of the hologram can be read

with a portable optical reader equipped with

Bluetooth®. These sensors exhibit long-term

chemical and physical stability, enabling

measurements over long time periods without

evidence of hysteresis. A further advantage is the

ability of holographic analyte sensors to be

incorporated into a multitude of formats (e.g.

catheters, contact lenses, implants), thereby offering

increased patient choice.

The prototype hand-held device (figure 3) has a

touch-screen user interface. The reader employs a

bespoke spectrometer optical sub-assembly,

combined with analogue to digital converters

(ADCs). These ADCs are read by a microprocessor

which provides a calibrated output of wavelength

and the corresponding concentration of the analyte

being measured. This is achieved by use of look up

tables or calibrated polynomial curve fit data.

Temporal data trends can be calculated on-device or

data can be sent via Bluetooth® to a HH web service

to enable downstream decision support.

3.2 Biometric Authentication

The HH system design has evaluated scenarios

where people entering data might be easily confused

by similar names or as different members of the

same family. A number of protocols for

disambiguation and absolute patient identification

can be incorporated, including biometrics and / or

smartcards. Biometric variables can be classified

either as physiological (e.g., derived from a

fingerprint, face or iris scan) or behavioural (e.g.,

speech recognition) (Biometrics, 2007).

The HH system requires technology options that

are cost-effective, fast and accurate. Based on these

criteria, biometric identification used robust

fingerprint recognition technology (‘2’ in figure 1)

(Wilson 2003). For flexibility, the biometrics device

was designed to be “loosely-coupled” with the

sensor and the home hub modules (refer to section

3.3). This permitted different packaging options

with integration of new sensor types without

substantial additional engineering cost. The hub is

used to tag the biometric ID to the incoming sensor

data in order to form an association between a

patient identity and a data reading. Communication

between sensor, home hub and the biometrics

module is via Bluetooth®, using a protocol specified

by IBM, Zarlink and Smart Holograms.

In its current operational mode, the biometric

device (figure 4) used in the HH system saves the

scanned fingerprint of a user in local memory, and

the individual is prompted to assign an “ID” specific

to that fingerprint. That ID is then transmitted via

Figure 2: Real-time measurement of blood glucose using a

holographic sensor compared to measurements made off-

line and post experiment by a traditional reference

method.

Figure 3: Smart Holograms prototype Hologram reader

with integrated Bluetooth

connectivity.

HEALTHINF 2008 - International Conference on Health Informatics

122

Bluetooth® to the data hub to be tagged within an

electronic patient record. This “enrolment” process

can be used prior to sensor readings to validate

identity. A delay-free smartcard that has high end-

user acceptance (e.g. contact-less ‘wave & pay’

cards) can also be used in appropriate circumstances

e.g. to enable inter-service access along the patient

path. Identification of the patient by means of their

enrolled ID sends that ID to the patient-proximal

hub (in the clinical data ‘baseline data’ operational

hub or the patient’s own home hub, or mobile hub).

The home hub can associate the subsequent sensor

reading with the patient ID. This is the

“identification” process. The design of the device

ensures security of patient information. All scanned

fingerprints are stored in the memory of the

biometric device. No fingerprint scans are

transmitted. All patient-identifiable information can

be encrypted at source, in transmission and storage.

If a hacker were to capture the wireless

transmissions the data would have no meaning.

3.3 Home Hub

The function of the ‘home hub’ (‘3’ in figure 1) is to

collect and collate the data from sensor(s) and the

biometric and smartcard / reader device(s) and to co-

transmit these via an appropriate communication

channel to the remotely-located server. The home

hub sits at the centre of the data collection and

transmission capability of the system, where all

devices - sensors, authentication module and server

connect through common interfaces. Various

physical realizations of the hub are possible. It can,

for example, be a mobile device such as a standard

mobile phone (we have demonstrated functionality

on a Sony Ericsson P910 phone - figure 5) that can

provide near real-time data connectivity. The hub

can also be deployed as a fixed ‘wired’ hub using,

for example, Ethernet connectivity. In a clinical

environment this also provides near real-time data

connectivity. For home use, hubs can be configured

to upload data periodically e.g. once or twice a day.

Figure 5: Mobile phone personal data hub by IBM.

Sensor devices currently connect to the hub

through adapters specific to the type of interface

required. Adapters are device transmission protocol

specific and can be developed by any manufacturer

wanting to provide connectivity of their devices to

this infrastructure. In HH, the medical devices use

Bluetooth®, although the architecture allows for this

to be any available. Zigbee has some power

consumption advantages over Bluetooth

(Zigbee,

2007).

We have developed adapters for a variety of

devices that can be used in conjunction with the

demonstrator system; namely for: (1) weighing

scales; (2) blood pressure cuff; (3) pulse oximeter;

(4) glucose meter. In addition to adapters for sensor

devices, an adapter is also required to interface the

hub to an appropriate application server. This step is

also protocol-specific and in practice a range of

adapters may be needed according to specific

application scenarios.

The hub architecture (figure 6) consists of an

event engine that behaves much like a broker, in that

it can receive events from one adapter and passes

these onto another adapter. In the simplest operation,

the event engine receives events from a single sensor

device (for example, weighing scales) and sends this

event to the adapter that transmits these events to an

application server. The current implementation of

this architecture uses the IBM Personal Care

Connect toolkit (Blount, 2007) and is based around a

number of standard technologies. These include:

• Java 2, Micro Edition, Mobile Information

Device Profile, Connection-Limited Device

Configuration (J2ME, MIDP, CLDC) or OSGI

for fixed hub (Java ME, 2007).

Figure 4: Fingerprint scanner by Zarlink.

NOVEL SENSOR TECHNOLOGY INTEGRATION FOR OUTCOME-BASED RISK ANALYSIS IN DIABETES

123

• JSR-82 to allow Java to interface to Bluetooth®

(assuming Bluetooth-enabled medical devices)

(JSR 82, 2007).

• Bluetooth® or mobile connectivity via GSM /

GPRS for wireless hubs.

• Ethernet, ISDN, etc. for fixed / wired hubs.

The hub design supports multiple adapters – key

to creation of an open-hub platform. Each adapter is

specific to a hub family, which is characterised by

the following attributes: (i) device-to-hub

communication, (ii) event data representation, (iii)

event serialisation, and (iv) hub-to-server

interaction. Different hub families can use different

event classes, communication protocols, and

serialisation schemes. A hub family may be

optimised to handle a specific set of biomedical

sensor types. Each adapter handles communication

with a family of hubs by (i) defining a protocol, (ii)

de-serialising the data forwarded by the hub, and

(iii) extracting event data from the event after

instantiating the events. Adapters also convert

proprietary data events into a consistent format – i.e.

act as a definable interface. It is expected that HH

will migrate to industry recommendations of

consortia such as the Continua Alliance (Continua,

2007). The hub functionality is currently

implemented in Java and is deployable to a mobile

phone as a MID-let or to a fixed hub as a set of

OSGI bundles.

In the current research demonstrator, sensor data

can be obtained from devices and stored on the hub

prior to transfer by a standard HTTP protocol. A

HTTP ‘Post’ agent uses stable storage to cache

events that cannot be sent immediately and then

forwards them when the next connection to the

server is established. Data sent to the server is

secured using the Secure Sockets Layer (now

referred to as Transport Layer Security) (SSL, TLS).

Other options include: Access Point Node (APN)

and Virtual Private Network (VPN).

Secure Sockets Layer: SSL is commonly used

within internet applications to provide secure client

to web server connections.

Access Point Node: The Access Point Node (or

Name) is the definition of the internet connection on

a GPRS mobile device that provides the route the

data will take from the GPRS device to access other

networks such as the Internet. All APN’s are defined

within the mobile device with a username and

password. The APN concept assumes the existence

of GPRS support nodes that exist between a mobile

device (using a GSM or UMTS service) and a server

providing IP-based access to the Internet. By

obtaining a private APN (with its own unique

credentials) it is possible to create a private

connection back to the HH servers. The HH

demonstrator has been successfully tested with APN.

Virtual Private Network: A VPN effectively

creates a private network by creating a secure

‘tunnel’ through the existing IP network. The HH

demonstrator has successfully used the IBM VPN

product WebSphere Everyplace Connection

Manager (WECM).

A sub-class of specialised functions that mimic

adapters exist to create functions internal to the hub.

A number of these have been implemented within

the current demonstrator. For example, an “Audio

Alert Agent” makes pre-configured audio alerts in

response to certain events like the reception of data

from a device or successful transmission of an event

to the server. The “User Display Agent” enables the

display of status information and provides input

capability for the user.

3.4 Server-side Architecture

The server-side architecture comprises of a ‘QUID’

and ‘QUIRA’ component (figure 7 and ‘4’ in figure

1) and has been described in detail elsewhere

(Subramanian, 2006, Shaikh-Ali, 2007, Conley,

2007).

The QUID component focuses on delivering data

collection, data storage, process execution and portal

Figure 6: Hub technical overview.

HEALTHINF 2008 - International Conference on Health Informatics

124

infrastructures to support users of the system

including clinicians, patients and researcher roles.

The development of QUID has been guided by the

requirements of Integrated Care Pathways (ICPs) for

diabetes (ICP, 2007). The QUID component

collects, validates and stores data streams from hubs

along the patient path. The interface for data

presentation / collection emulates workflows in real-

world care pathways. This ‘end-user familiarity’

design feature is strengthened by presentation of the

sensor data in a timeline-based (longitudinal)

layered manner, make the data meaningful. We

have used OpenLaszlo (Laszlo, 2007) to display

charts in an intuitive manner. Similarly, the patient

portal permits access to personal data supporting the

self-management paradigm

The QUIRA (Quantitative Individualised Risk

Analysis) component comprises a risk analysis

engine performing various operations identifying

signals in the longitudinal data stream, alerting the

care team to fulfilment of pre-defined risk criteria.

QUIRA represents ongoing research we will report

on in the future.

4 CONCERNS & FUTURE WORK

“Pervasive” or “ubiquitous” computing covers a

range of research topics, including distributed

computing, mobile computing, sensor networks,

communications, artificial intelligence, and human-

computer interaction. It is an emerging field of

research, and as such has many unresolved issues –

notably in areas like security, usability, privacy and

ethics amongst others. In the framework described in

this paper, ubiquitous computing devices demand

secure transmission of data to the server, in turn

demanding encryption mechanisms that defeat

purposeful or accidental ‘eavesdropping’. Device

miniaturisation assists the resolution of powering

issues. Significant ethical and privacy issues remain,

and it is axiomatic that a comprehensive informed

consenting process needs to be developed that is fit-

for-the defined purpose of disease early detection

and prevention. Informed consent needs to be

properly structured in the recording workflow.

There is currently no widely-accepted standard

protocol for the device-to-server data transmission

and / or format / structure for data being transferred

between devices. We expect interoperability

between devices from different vendors to be a key

focus within the scope of the Continua Healthcare

Alliance (Continua, 2007). Irrespective of global

technical standards adopted, methodological

standardisation of data acquisition needs to be

defined in order for patients and carers to reap the

benefits of interoperable systems.

The HH project has to date employed non-

invasive sensors. However we anticipate significant

developments in coming years in the area of

invasive (implantable) sensors. Implantable sensors

will likely be micro-miniaturised devices that can be

implanted into a patient’s body to enable relaying of

health-critical signals on a semi-continuous basis.

5 CONCLUSIONS

System-based management of chronic conditions is

essential to improve healthcare outcomes.

Conventional models of healthcare provision lack

capacity to continuously monitor physiological data

Figure 7: Physical implementation of the conceptual design shown in Figure1.

NOVEL SENSOR TECHNOLOGY INTEGRATION FOR OUTCOME-BASED RISK ANALYSIS IN DIABETES

125

combined with life event ‘timelining’. In such an

information system, healthcare can be provided in a

‘patient-centric’ model that maximises healthcare

resources.

ACKNOWLEDGEMENTS

This project has been funded by the Inter Enterprise

Computing Theme of the UK Department of Trade

and Industry (DTI)-led Technology Programme, for

which we are grateful. ECC and DRO are grateful

for support by the Wales Office of Research and

Development for Health and Social Care, Wales

Assembly Government.

REFERENCES

Abidi, S.S.R, 2001, “An Intelligent Tele-Healthcare

Environment Offering Person-Centric and Wellness-

Maintenance Services”, Journal of Medical Systems,

Vol. 25, No.3.

Biometrics, a, date visited: July 2007, link visited:

http://en.wikipedia.org/wiki/Biometrics

Blount, M., Batra, V.M., Capella, A.N., Ebling, M.R.,

Jerome, W.F., Martin, S. M., Nidd, M., and Wright,

S.P., 2007, “Remote health-care monitoring using

Personal Care Connect”, Volume 46, Information-

Based Medicine, IBM Systems Journal.

Conley, E.C., Owens, D.R. Luzio, S., Subramanian, M.,

Shaikh Ali, S., Hardisty, A., Rana, O. (2007)

“Simultaneous Trend Analysis for Evaluating

Outcomes in Patient-Centred Health Monitoring

Services”, Springer Journal of Healthcare

Management Science, Submitted.

Clemensen, J., Larsen, S. B., and Bardram, J., 2004,

“Developing Pervasive e-Health for Moving Experts

from Hospital to Home”, Proceedings of the IADIS e-

Society Conference, pp.441-448, Avilla, Spain.

Common User Interface, June 2007, link visited:

https://www.cui.nhs.uk/Pages/NHSCommonUserInterf

ace.aspx

Continua Healthcare Alliance, June 2007, link visited:

www.continuaalliance.org

Crossbow, June 2007, link visited: www.xbow.com

FIPS, 2002. “Security Requirements for Cryptographic

Modules”, link accessed

http://csrc.nist.gov/cryptval/140-2.htm

Gao, T., Massey, T., Sharp, J., Bishop, W., Bernstein, D.,

and Alm, A.., September 2006 “Integration of Triage

and Biomedical Devices for Continuous, Real-Time,

Automated Patient Monitoring”. Proceedings of the

IEEE Medical Devices and Biosensors Conference

(IEEE MDBS 2006), Boston, MA..

Hein, A., Nee, O., Willemsen, D., Scheffold, T., Dogac,

A., Laleci, G., 2006, “SAPHIRE - Intelligent

Healthcare Monitoring based on Semantic

Interoperability Platform - The Homecare Scenario”,

1st European Conference on eHealth (ECEH’06),

Fribourg, Switzerland.

Java ME: Mobile Information Device Profile (MIDP), date

visited: July 2007, link visited:

http://java.sun.com/products/midp/

JSR 82: Java APIs for Bluetooth, link visited:

http://www.jcp.org/en/jsr/detail?id=82

King, H., Aubert, R., Herman, W., 1998, “Global burden

of diabetes, 1995-2025: prevalence, numerical

estimates, and projections” Journal: Diabetes

Care,Volume: 21 Issue: 9, pp 1414-31.

Lorincz, K., Malan, M., Fulford-Jones, T., Nawoj, A.,

Clavel, A., Shnayder, V., Mainland, , S., Welsh, M.,

Oct-Dec 2004, “Sensor Networks for Emergency

Response: Challenges and Opportunities”, In IEEE

Pervasive Computing, Special Issue on Pervasive

Computing for First Response.

Laszlo, date visited: July 2007, link visited:

http://www.laszlosystems.com/.

ICP, Diabetes NSF (Wales) Standards, link accessed:

www.wales.nhs.uk/sites3/home.cfm?orgid=440

Shnayder, V., Chen, B., Lorincz, K., Jones, T., & Welsh,

M., 2005, “Sensor Networks for Medical Care”,

Technical Report TR-08-05, Division of Engineering

and Applied Sciences, Harvard University.

Shaikh-Ali, A., Rana, O.F. Hardisty, A., Subramanian, M.,

Luzio, S., Owens, D.R. and Conley, E.C., October

2007, "Portal Technologies for Patient-centred

Integrated Care", European Conference on eHealth,

Oldenburg, Germany.

SAPHE project, July 2007, link visited:

http://ubimon.doc.ic.ac.uk/saphe/m338.html,

Subramanian, M., Shaikh-Ali, A., Rana, O.F. Hardisty, A.

and Conley, E.C., October 2006, "Healthcare@Home:

Research Models for atient-Centred Healthcare

Services", John Vincent Atanasoff International

Symposium on Modern Computing, Sofia, Bulgaria.

IEEE Computer Society Press.

WHO, World Health Organisation, date visited: July 2007,

link visited: http://www.who.int/en/

Wilson, C., Hicklin, R.A., Korves, H., Ulery, B., Zoepfl,

M., Bone, M., Grother, P., Micheals, R., Otto, S.,

Watson, C., 2003, “Fingerprint Vendor Technology

Evaluation”, National Institute of Standards and

Technology, Mitretek Systems, NAVSEA Crane

Division

Zigbee, date visited: July 2007, link visited:

http://www.zigbee.org/en/index.asp

HEALTHINF 2008 - International Conference on Health Informatics

126