Journal of Epidemiology and Global Health (2012) 2, 1–13

http : / / www.elsev ier .com/ locate / jegh

Can pervasive sensing address current challengesin global healthcare?

Louis Atallah, Benny Lo, Guang-Zhong Yang *

Hamlyn Centre, Imperial College, London SW7 2AZ, UK

Received 5 July 2011; received in revised form 23 November 2011; accepted 25 November 2011Available online 30 January 2012

02do

fa

do

Th

KEYWORDSGlobal healthcare;Technology;Pervasive sensing;Ageing;Chronic disease

63-2373/$ - see front mai:10.1016/j.jegh.2011.11

* Corresponding author.x: +44 2075945196.E-mail addresses: latallac.ic.ac.uk (B. Lo), g.z.ya

is is an open access articl

tter ª 201.005

Tel.: +44

h@doc.ic.ng@imperi

e under the

Abstract Important challenges facing global healthcare include the increase in thenumber of people affected by escalating healthcare costs, chronic and infectiousdiseases, the need for better and more affordable elderly care and expanding urban-isation combined with air and water pollution. Recent advances in pervasive sensingtechnologies have led to miniaturised sensor networks that can be worn or inte-grated within the living environment without affecting a person�s daily patterns.These sensors promise to change healthcare from snapshot measurements of phys-iological parameters to continuous monitoring enabling clinicians to provide guid-ance on a daily basis. This article surveys several of the solutions provided bythese sensor platforms from elderly care to neonatal monitoring and environmentalmapping. Some of the opportunities available and the challenges facing the adoptionof such technologies in large-scale epidemiological studies are also discussed.ª 2012 Ministry of Health, Saudi Arabia. Published by Elsevier Ltd. This is an openaccess article under the CC BY-NC-ND license (http://creativecommons.org/licenses/

by-nc-nd/4.0/).

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22. Global challenges in healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1. Chronic and infectious diseases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2. Ageing populations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3. Urbanisation and environmental impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.4. Rising costs of healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.5. Prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3. How can pervasive sensing address these challenges? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.1. Pervasive sensing for infectious and chronic diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.2. Understanding ageing and research on elderly care . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2

2075940809/2075941499;

ac.uk (L. Atallah), benlo@al.ac.uk (G.-Z. Yang).

Ministry of Health, Saudi Arabia. Published by Elsevier Ltd.CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

2 L. Atallah et al.

3.2.1. Beyond telecare and enabling independent living. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63.2.2. Fall detection and management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.3. Understanding the influence of environmental factors on health . . . . . . . . . . . . . . . . . . . . . . . . . . 63.4. Innovations in child and maternal health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.5. Applications in epidemiological studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4. A Common ground for innovation in pervasive healthcare. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.1. The role of communication technology and mobile networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.2. Adopting pervasive sensing technologies in healthcare and wellbeing . . . . . . . . . . . . . . . . . . . . . . . 84.3. Technical challenges to using pervasive sensors for healthcare scenarios . . . . . . . . . . . . . . . . . . . . 94.4. Challenges in market uptake of pervasive sensing solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104.5. Challenges in adopting pervasive sensing solutions in large epidemiological studies . . . . . . . . . . . . . 104.6. Emerging economies and changes in healthcare policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104.7. Opportunities in pervasive sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1. Introduction

Several trends are influencing global healthcareleading to rapid changes in both the developmentand adoption of new technology. These trends needto be studied carefully whenever new healthpolicies are introduced, especially for developingmarkets in both developed and developing coun-tries. Escalating costs of healthcare highlight theimportance of managing resources with a betterfocus on system performance and patient well-being. Urbanisation closely linked to industrialisa-tion means the dramatic rise of city populationsworldwide, especially in developing countries, lead-ing to higher rates of many blood-born, infectiousand pollution-related diseases [1]. The access tobetter treatments for patients in urban areas, onthe other hand, leads to a drop in several otherareas. Coupled with this, in almost every countryin the world, the proportion of people aged over60 is growing faster than all other age groups.Although this can be seen as a success of modernhealthcare, it challenges the society to adaptquickly in order to improve life-long health andwell-being and overall health of the elderly, en-abling them to lead an independent and active life.

Consumer-driven healthcare plans in the devel-oped world, combined with wireless/mobile solu-tions and a growing use of web-based platformshave enabled better access to healthcare data forboth patients and clinicians. However, these solu-tions highlight their own problems of scalability,security and privacy. Despite significant advancesinmedical devices, currentmonitoring tools are onlyable to provide a snapshot of a person�s physicalparameters rather than offering a continuous viewof his/her wellbeing. In most cases, detailed moni-toring only takes place during a person�s stay in ahospital or clinic. Recent advances in sensors and

embedded computing technologies have led tominiaturised sensor networks that can be worn orintegrated within the home environment withoutaffecting the person�s lifestyle or daily activities.Fig. 1 shows examples of the evolution of computersfrom large to miniaturised wearable and implant-able devices. Integrated micro-sensors with on-board processing and wireless data transfer are thebasic components of such networks already inexistence. Extensive research has been directed tobio-sensor design, interfacing and bio-compatibil-ity, wireless communication, ultra-low power pro-cessing, power scavenging, autonomic sensing, andintegrated wireless sensor micro-systems leadingto pervasive systems. Using the body as the mediaand a source of inspiration, energy to provide long-term continuous sensing and monitoring is the maindriving force behind the current development ofbody sensor networks (BSN) [2]. These sensors haveprovided new means of addressing the followingchallenges tackling some of the global healthchallenges:

• Remote and continuous monitoring of the progress ofchronic and infectious diseases.

• Observing post-operative patients at home allowingearlier release from hospital, reducing costs and pro-viding efficient 24-hour monitoring.

• Understanding ageing and providing novel techniquesfor elderly care management.

• Large-scale monitoring of environmental changes andthe impact of urbanisation.

• Providing innovative techniques for maternal and neo-natal care after birth.

This article will highlight several challengesfacing global healthcare, then focus on pervasivetechnologies that can address some of the chal-lenges posed. Finally, some of the issues thatmay arise when new technologies are adopted that

Fig. 1 Examples of the evolution of computers over time from large computers to wearable and implantable devices.

Can pervasive sensing address current challenges in global healthcare? 3

influence patient health and well-being will be dis-cussed, as well as opportunities for the applicationof pervasive sensing within epidemiological stud-ies. Table 1 summarises some of the challengesdiscussed, their requirements and the sensingsolutions proposed.

2. Global challenges in healthcare

2.1. Chronic and infectious diseases

Chronic diseases, including heart disease, cancer,respiratory diseases and diabetes are the leadingcause of death worldwide. The major risk factorsassociated with these diseases include smoking,alcohol, physical inactivity, obesity and a poordiet. By eliminating major risks, it is expected thataround 75% of heart disease, stroke and type 2 dia-betes and 40% of cancer would be prevented.1 Inaddition to chronic diseases, infectious diseasesare also a significant burden on global economiesand public health. Emerging infectious diseaseevents have risen significantly and many eventsare dominated by zoonosis, and are influenced bysocio-economic, environmental and ecologicalfactors. A study by Jones et al. [3] concluded thatglobal resources to counter disease emergence arepoorly allocated, with the majority of the scientificand surveillance effort focused on countries fromwhere the next important emerging infectious dis-ease is least likely to originate. This highlights theneed for coordinated global approaches to prevent-ing and controlling complex infectious diseaseproblems.

1 http://www.who.int

2.2. Ageing populations

The demographic shift associated with the ageingpopulation is another major healthcare challenge.The United Nations world population report high-lights the current rates of population ageing whichare unprecedented in human history. This is mainlycaused by declining fertility and mortality rates.The young/old balance is shifting throughout theworld. Although the proportion of octogenariansis higher in more developed regions, most countriesshare an increase in old-age dependency ratiowhere the elderly are supported by a relativelysmaller number of potential contributors. Labour-force participation among the older populationhas been declining worldwide, although the ratesremain higher in the less developed than the moredeveloped countries. The rapid increase in elderlyrates builds a strain on most existing systems, sodo diseases that are considered typical of elderlypopulations, including osteoarthritis, diabetes mel-litus and hypertension. Elderly people are also gen-erally susceptible to bronchitis, osteoporosis andheart disease. Common risk factors are obesity,diabetes, hypertension and various cardiovasculardisorders [4].

Encouraging the elderly to continue to partici-pate in their societies with dignity and security, de-spite having chronic conditions in some cases, is achallenge which is faced by healthcare systemsand policy makers worldwide. Another challengeis that of falls which are the leading cause of injuryand death among the elderly [5]; 4 out of 7 falls arefatal in persons over 70. Post-fall injuries can trig-ger a rapid decline in health and increase the riskof elderly death. Thus, fall-risk assessment beforea fall occurs could be of great help in avoiding suchscenarios.

Table 1 Several challenges facing global healthcare, requirements and solutions that can be provided by pervasivesensing.

Global health challenge Requirements Current systems Pervasive healthcare

Chronic diseasesPost-operative recoveryActivity measurement

PhysiologicalparametersActivity measurement/feedbackSleep assessment

Snapshot measurementSurveys/observationsPersonal feedback/surveys

Continuous monitoringActivity/energyexpenditurePervasive sensors forsleep qualityIntelligent data miningover time

Falls and elderly care Fall detection/fall riskassessment

Pendants with push-buttonsClinic visits/surveysClinical observations

Automatic fall detectionWearable sensors forbalance/stability

Rehabilitation Encouraging activityMeasuring performance

Attending rehab clinicsClinical observation

Remote rehabilitationmonitoringSensors for quantifyingperformance

Maternal and neonatalcare

24-Hour monitoring ofneonates

Clunky uncomfortablesensors

Miniaturised sensorsembedded in clothes/surroundingsMinimal disruption toneonates

Environmental monitoring Long-term monitoringDetecting changes

Satellite/radar systemsAutomatic imageanalysis

Miniaturised environ-mental sensorsIntelligent eventdetection

4 L. Atallah et al.

2.3. Urbanisation and environmental impact

Most of the world�s population now lives in cities. Intheory, living closer to health providers in citiesshould lead to better access to medications andtreatments. However, this is not always the casein many countries around the world. The link be-tween urbanisation and health problems is perhapsmore evident in developing countries like India,China and Brazil where changes are occurring atan unprecedented pace. Cities are becoming hubsfor infectious disease transmission, mainly becauseof high population densities as evidenced by recentdisease pandemics [6]. Rapid urbanisation also in-creases mortality rates from diseases linked to life-style changes, such as cancer and cardiovasculardiseases [7].

Many environmental changes are also linked tourbanisation. The WHO report on sanitation anddrinking water2 shows that 137 million city dwellersworldwide have no access to safe drinking waterand more than 600 million do not have adequatesanitation. These factors are prime contributorsto the spread of infectious diseases, such as intes-tinal parasites leading to diarrheal infections andmight cause cholera endemicity [6]. Air pollution

2 http://www.who.int

has an impact on mortality with an association withcardiopulmonary and lung cancer [8] and an effecton the body�s respiratory and cardiovascular sys-tem. Health effects of air pollution also includebreathing difficulties and worsening of existingchronic conditions such as asthma leading to in-creased medication use, hospital admissions andpremature mortality.

2.4. Rising costs of healthcare

Rising healthcare cost is now a worldwide phenom-enon to which many factors have contributed thusfar. The most important being the prevalence ofchronic and infectious diseases as mentionedabove, in addition to providing long-term care forpatients [9]. Another is the availability of noveltechnologies and the demand for them, especiallyin emerging economies. Rising hospital costs com-bined with increased patient demands are alsodriving factors. Obesity, especially in developedcountries [10], has an impact on medical spending,with an estimated increase of 12% owing to obesityalone in some countries. Longevity brings the chal-lenge of providing healthcare for a larger numberof people with changing demographics. Rising la-bour, medication and energy costs over time arealso contributing factors.

Table 2 Some chronic diseases, physiological parameters that are of clinical importance and possible sensors that can beused to observe them.

Disease Physiological parameters Possible sensors

COPD Respiration, heart rate Wearable heart rate/respiration sensorsOxygen saturation and activity levels Accelerometers

Parkinson�s disease Gait, muscle tone, activity Accelerometers, optical/vision sensorsGyroscopes

Hypertension Blood pressure/activity Wearable blood pressure sensorAccelerometers

Cancer Weight loss, activity, behaviour patterns Accelerometers, gyroscopesWeight sensors

Arthritis Gait patterns, temperature, stiffness Accelerometers, optical/vision sensorsGyroscopes

Diabetes Gait patterns, visual and sensory impairment Accelerometers, vision, glucose monitorsGyroscopes

Cardiac arrhythmia Heart rate, ECG ECG sensors

Heart failure Blood pressure Heart rate and blood pressure sensors

3 http://www.chronious.eu4 http://www.wasp-project.org5 http://heartcycle.med.auth.gr6 http://www.hitech-projects.com/euprojects/myheart7 http://saal.ie/projects8 http://www.bosch-telehealth.com

Can pervasive sensing address current challenges in global healthcare? 5

2.5. Prevention

Several debilitating and costly diseases, such ascardiovascular disease, cancer, diabetes and respi-ratory diseases are linked by several risk factors.Hypertension, smoking, cholesterol, malnutritionand poor diets, obesity, alcohol and physicalinactivity are highlighted as the leading causesfor preventable deaths worldwide [11]. The WHOindicates that most health-care systems are basedon responding to acute problems and urgent needsof patients rather than preventative care. Althougha shift toward preventative measures is seen insome countries, risk-reducing behaviours shouldbe encouraged, especially for vulnerable groups.

3. How can pervasive sensing addressthese challenges?

Pervasive sensing is increasingly becoming a part ofeveryday healthcare owing to advances in sensortechnology, its increasing availability, continuingminiaturisation and cost reduction [12,13]. Illus-trated below is how pervasive sensing technologiescan provide potential solutions to some of theaforementioned global challenges.

3.1. Pervasive sensing for infectious andchronic diseases

Early detection of pathogens is crucial in the treat-ment and prevention of infectious disease. Nano-scale materials with unique optical and magneticproperties have been efficiently used for the

detection of pathogens as summarised in the sur-vey by Tallury et al. [14]. Nanodiagonistics usenano-technology in clinical diagnosis of severalinfectious pathogens, including Escherichia coli,Salmonella and Influenza. Although these methodsare now miniaturised, they are still non-pervasiveas they normally require expert analysis and equip-ment. Further, to correct and to provide rapid-diagnosis, pervasive sensing can be used to observepatients recovering from a certain infectiousdisease, in terms of both their physiological param-eters and behaviour patterns.

Research into chronic disease management, onthe other hand, has shown several promising sens-ing systems for monitoring different parametersover time. Recent projects that are investigatingor have investigated the use of sensors for chronicdisease management include the European pro-jects Chronious,3 WASP,4 HeartCycle,5 MyHeart6

and SAAL.7 Table 2, based on information pre-sented in Body Sensor Networks [2] summarisessome of the sensors for the observation of parame-ters relating to some of the most common chronicdiseases. Even simple sensing paradigms wherepatients can use devices to remotely record andsend sensor measurements to their clinicians, usinga telecare device such as the Health Buddy,8 forexample, have proven to be efficient for observing

6 L. Atallah et al.

diseases such as asthma [15], diabetes and heartfailure. One of the main challenges now is to makethese sensors pervasive enough and embed them inthe patient�s surroundings without affecting dailypatterns and routines.

For COPD (Chronic obstructive pulmonary dis-ease), for example, a positive association existsbetween training intensity during rehabilitationand maximal oxygen consumption and musclestrength [16]. Thus, several of the previous workson COPD have been on activity detection and itscorrelation with recovery. This also applies to car-diac rehabilitation and post-operative monitoring.

3.2. Understanding ageing and research onelderly care

Ageing, being one of the most important challengesfaced by both developing countries and the devel-oped world, has led to several research directionsfocussing on how pervasive sensing can helpresearchers understand the ageing process itself,as well as dealing with some of the aspects ofageing. The following sections summarise theseresearch directions.

3.2.1. Beyond telecare and enablingindependent livingThe review conducted by Barlow et al. [17] investi-gated the use of telecare for frail, elderly peopleand those with chronic conditions, summarisingthe results of 8666 studies. Based on the study,the most effective telecare interventions appearedto be automated vital signs monitoring (for reduc-ing health service use) and telephone follow-upby nurses. The cost-effectiveness of these inter-ventions was less certain.

The aim of telecare is to provide the elderly withmeans of conducting their daily activities withminimal or no intervention. Combining telecaresystems with wearable and ambient systems canenable continuous care as well as the observationof changes in behaviour routines [18,19]. Severalstudies have been conducted worldwide for thecombination of sensors with telecare, and manysystems are summarised in ‘‘A review of approachesto mobility telemonitoring of the elderly in theirliving environment’’ [20]. The sensors used in-cluded passive infrared (PIR) sensors, magneticsensors observing door use, bed-sensors, sound sen-sors, electrical-usage sensors, wearable sensorsincluding accelerometers, gyroscopes and magne-tometers or a combination of several wearableand ambient sensing modalities [18,21]. In additionto monitoring, several commercial telecare solu-tions include a digital alarm unit that can either

be triggered by pushing a button or by a suddenchange in vital signs or a fall.

3.2.2. Fall detection and managementGiven that falls are the leading cause for injuryamong the elderly, a large amount of pervasivesensing research has focused on fall detection.One of the most dangerous side effects of a fall isthe prolonged period the elderly spend on theground and the risk factors attached thereto. Thus,early detection and rapid response by trained staffis essential in fall management. Several pendantsystems for fall-reporting are available, but requirethat the elderly press a button after a fall. Thismight not be possible in case they suffer from a lossof consciousness. Automatic methods for falldetection using sensing include the use of image-based sensing [22], arrays of acoustic detectors[23], gyroscopes [24], accelerometers on theirown or combined with air-pressure sensors [25].

Several strategies can be adopted for the pre-vention of falls, including specific exercises, edu-cation and assessing homes for extrinsic factors.Medical assessment can also identify intrinsiccauses of falls and is normally based on a clinician�sobservation of patient performance during a lab-based test. Recently, pervasive sensing has beenused to assess the risk of falls aimed at providinga method of continuous assessment without theneed for a hospital visit. Examples are the workby King et al. using an ear-worn accelerometry de-vice [26], wearable shoe-insoles by Oberzaucheret al. [27] and waist-worn accelerometers byGietzelt et al. [28].

3.3. Understanding the influence ofenvironmental factors on health

Environment and climate changes have alreadybeen monitored constantly by meteorologists, geol-ogists, environmentalists, biologists and biochem-ists, among others. This is mostly done throughusing remote sensing technologies, such as radaror satellite images, taking soil and water samplesand installing air pollutant sensors in urban areas.However, owing to the extensive cost of traditionalsensing technologies, such as satellite imaging, onlymacro measurements around large regions can becaptured, and often only remedy actions can be ta-ken in case of environmental incidents after theyhave occurred. In addition, the direct influence ofenvironmental factors on human health is often verydifficult to identify.

The recent advances in ultra-low power and low-cost wireless sensing technologies can significantlyimprove environment monitoring and enable

Can pervasive sensing address current challenges in global healthcare? 7

real-time monitoring of micro-environments. Re-cent studies have demonstrated the feasibilityand effectiveness of applying wireless sensornetwork (WSN) technologies to environment moni-toring. These studies include: automatically moni-toring soil conditions [29], the eruption of activevolcanoes [30]; the impact of wind farms in coastalareas; and the behaviour of icecaps and glaciers[31]. Pervasive sensors also enable the monitoringof water supply systems [32] and air pollution mon-itoring [33]. With detailed environmental andhealth information acquired using the new wirelessenvironmental sensors and pervasive health sen-sors, direct environmental impact on human healthcan be identified accurately. It will not only betterthe understanding of environmental health, but itwill also help introduce precautionary measure-ments to minimise the effect of environmentalhazards on human health.

3.4. Innovations in child and maternal health

The observation of physiological parameters,including respiration, heart rate and oxygen satura-tion (SO2) is crucial to the monitoring of prematureor critically ill neonates. Continuous care is neededfor the early detection of medical problems such asarrhythmias, convulsions and apnea. Adhesive andlarge, tethered sensors can cause discomfort tobabies with fragile skin. The advances in wirelesssensing technology have allowed the miniaturisa-tion of the sensors used and their embedding inwearable garments, such as a jacket for neonatalECG monitoring [34] and a biosensor belt forobtaining heart rate, breathing rate, movementand temperature [35]. A wearable wireless pulseoximetry oxygen saturation (SpO2) device usingreflectance sensors was proposed in ‘‘Non-invasiveblood oxygen saturation monitoring’’ [36] to obtainheart rate and oxygen saturation.

3.5. Applications in epidemiological studies

The previous sections presented a summary of howsensors have been integrated in different health-care applications ranging from chronic diseases tochild and maternal health. In this section, someof the current work aimed at using sensing inlarge-scale epidemiology studies will be surveyed.

Glucose monitoring has been one of the success-ful applications of pervasive remote sensing, espe-cially with the introduction of continuous glucosemonitors (CGM) that can help clinicians adjusttherapy according to changes in the condition ofthe patient [37]. Studies have shown that a CGMidentifies four times as many serious glucose excur-

sions as Self-Monitoring of Blood Glucose (SMBG).These sensors have been used for several patienttypes, such as children [38] and patients with type1 diabetes [39].

Another successful application of pervasivesensing has been that of using accelerometers forthe quantification of physical activity, as a pedom-eter for step count, or integrated in watches andwearable devices. The study by Troiano et al.[40], for example, observed 6329 participantswho provided at least 1 day of accelerometer dataand from 4867 participants who provided four ormore days of accelerometer data. These observa-tions allowed the authors to compare activity be-tween age groups, and deduce the adherence ofgroups to recommended amounts of physical activ-ity. Accelerometers were also used in a longitudi-nal study observing 1032 participants and indeducing the role of decreased activity in child-hood obesity [41].

Telemetry has been successfully used in severalstudies such as that of helping remotely observethe physical parameters/well-being of veterans(31,570 subjects [42]) and observing patients dur-ing their transport to hospitals in emergency med-ical service vehicles (494 patients) [43]. The nextsection will cover some of the challenges andobstacles facing the integration of pervasive sen-sors within such telecare systems.

4. A Common ground for innovation inpervasive healthcare

4.1. The role of communication technologyand mobile networks

The rapid growth of the communication technolo-gies and mobile networks have not only facilitatedthe information sharing for global health research,enhanced our understanding of diseases and influ-enzas, and improved the management of healthservices, but they have also enabled the realisationof pervasive health monitoring. With over 5 billionmobile phone connections, most of the world pop-ulation has access to the mobile network which canalso provide a backbone for pervasive sensingdevices and enable the capturing of invaluable per-sonalised health information. For instance, throughintegrating pervasive sensors, such as the BSN sen-sors [2], with mobile networks, detailed patientphysiological and activity information can be gath-ered among large populations. This would helpmonitor the effects of infectious and chronic dis-eases, providing enriched information for healthservices in improving disease management. Even

Fig. 2 Sensors that can be embedded in everyday surroundings: (a) the Sensewear armband; (b) ECG monitoring on acar seat; (c) a sensor-jacket for neonatal monitoring; (d) an ear worn activity recognition sensor; (e) the Nike+ipodshoe-based sensor for monitoring running; and (f) a sensor to observe wheelchair motion and paralympic athleteperformance.

8 L. Atallah et al.

simple mobile phone programs, such as the iStetho-scope,9 can potentially help with tracking the trendof chronic heart disease in the world population.

4.2. Adopting pervasive sensingtechnologies in healthcare and wellbeing

Fig. 2 shows several sensors that are currently usedpervasively to monitor different parameters,including: (a) the Sensewear armband to monitoractivity/energy expenditure10; (b) sensors embed-ded in a car seat to monitor heart function devel-oped by FORD and RWTH11; (c) a sensor jacketfor neonatal monitoring developed by Tu/e12; (d)the ear worn activity recognition sensor for gaitand activity monitoring developed by Imperial Col-lege London [44]; (e) the Nike+ipod that allows run-ners to use a shoe-based sensor to monitor runningperformance on a mobile phone or an ipod13; and(f) integrating a BSN node [2] on the wheel of awheelchair for observing wheelchair motion andparalympic athlete performance.14

9 http://www.guardian.co.uk/technology/2010/aug/30/iphone-replace-stethoscope10 http://sensewear.bodymedia.com/11 http://media.ford.com/news12 http://www.tue.nl/universiteit/faculteiten/industrial-design/onderzoek/onderzoeksprogrammas/di/projects/smart-jacket/13 http://www.apple.com/ipod/nike14 http://vip.doc.ic.ac.uk/esprit/

The survey by Cook et al. [21] provides a sum-mary of techniques deployed for providing ambientintelligence as a means of pervasive sensing withinthe context of smart homes and health monitoringand assistance. Ambient intelligence at home canbe used to support people with mental and physicalchallenges, helping them to lead independent livesby recognising activity [18], detecting changes oranomalies, reminding them of tasks [45], or assess-ing cognitive limitations.

Fig. 3 highlights several challenges facingglobal healthcare and the particular solutions thatcould be provided by sensing systems. These solu-tions include sensor fusion from both ambient andwearable sensors to provide personalised context-aware sensing for monitoring an individual�sbehaviour as well as the surrounding environment.An example of the use of sensing for continuoushealthcare is that of post-operative surgery moni-toring. Although current hospital recovery monitor-ing systems are effective during peri-operative andearly postoperative periods, they cannot be usedwhen the patient is at home. Following major sur-gery, patients are susceptible to complicationssuch as infections, abscess formation and malnutri-tion which can lead to slow recovery rates. In ‘‘Apervasive body sensor network’’ [44], for example,an ear-worn sensor was used to observe activitylevels of post-operative patients indicating changesover a few days after the operation. A wearable

Fig. 3 Several challenges facing global healthcare versus functionalities that could be provided by different sensingmodalities.

Can pervasive sensing address current challenges in global healthcare? 9

accelerometer was also used in ‘‘A new parameterfor assessing postoperative recovery’’ [46] for com-paring activity levels of patients undergoing laparo-scopic versus open gastrectomy. An importantaspect of post-operative recovery from differenttypes of operations, especially in the orthopaedicdomain, is the change in gait, and an amount ofwork has been dedicated to observing gait changesafter surgery using pervasive sensors. Accelerome-ters, gyroscopes and force-sensors have both beenused on different parts of the body to measure gaitparameters [47]. Force-sensitive insoles were usedto measure ground reaction forces and observe im-paired gait in ‘‘Real-time gait event detection’’[48]. Gyroscopes have also been used to observeknee-joint and hip-joint angles [49].

4.3. Technical challenges to using pervasivesensors for healthcare scenarios

From a technical point of view, although sensorsare now miniaturised and are equipped with effi-cient software, there are still challenges to beaddressed before weaving them into the fabric ofeveryday life. Energy optimisation plays an impor-tant role in using pervasive sensing for remotemonitoring. Energy consumption on sensor nodesdepends on battery type, sensor types used, radioand CPU usage. In order to balance between overallenergy consumption and performance, severalmeasures can be considered. These include theuse of data-compression, cache and proxy func-

tionalities which can improve performance byreducing the traffic as well as the data-load.Appropriate data management and event handlingcan also help maximise performance. Efficientwake-up methods (which can be added as specificservices) can save power and provide a rapidresponse to incoming events.

Reliability and dependability of sensor data (interms of access and accuracy) are both vital inorder to reduce false-alarm triggers, or moreimportantly to prevent loss of service which inhib-its detection of deterioration (in the patient�scondition) or a serious event (such as a fall) [50].Reliability involves having mechanisms for verifica-tion and correction as a guarantee of the reliablefunctioning of the BSN for long periods of time.The availability of several sensor types as well assensing frameworks highlights the problem of het-erogeneity in system deployment. The interopera-bility of these systems is of importance in makingsensing solutions useable in real-life scenarios. Thishas opened several research directions in terms ofproviding service-oriented architectures for theintegration of different types of sensors [51], aswell as proposing multi-agent systems that managethe infrastructure of services within the environ-ment both efficiently and securely by reasoning,task-planning, and synchronizing the data obtainedfrom the sensors [19]. Middleware solutions thatcan facilitate network functionality and deal withnode heterogeneity are also an area of research.The scope of the middleware used should cover

10 L. Atallah et al.

sensor nodes as well as devices and connected net-works. Examples include the open-source Uranusproject,15 the Cougar project16 and the SCADDSproject.17

Security constraints imposed by healthcareapplications have several important objectives,including confidentiality, integrity, traceabilityand trustworthiness. To make sure these objectivesare attained, several methods are used to safeguardwireless sensor networks from tampering, eaves-dropping and intrusion detection. These methodsinclude deploying secure software protocols forrouting, secure data aggregation and the securemanagement of groups of nodes.

4.4. Challenges in market uptake ofpervasive sensing solutions

Like most new technologies which are to beadopted in healthcare and the environment, perva-sive sensing technologies have a life-cycle to beconsidered: consulting with end users, testing insafe simulation environments, then real world tri-als after obtaining ethical/regulatory approval.

The success of these sensors will definitely de-pend on their level of miniaturisation, built-inintelligent analysis and most importantly theirintegration with everyday objects used by people.From a wearable sensor perspective, more sensorsare showing up embedded in textiles, watches, mo-bile phones, shoes and hearing aids. Ambient sen-sors, on the other hand, are all around, fromtemperature sensors to Passive Infra red (PIR) sen-sors and cameras that surround communities. It isalso worth noting that several sensors can be usedfor more than one modality. An example is theActiwatch,18 which includes time and activity mon-itoring, as well as sleep/wake patterns.

Ease of use of sensor nodes is of great impor-tance in deploying them in everyday applications.This is facilitated by ergonomic design and integra-tion within the surroundings, as well as wearableobjects, without affecting daily patterns of behav-iour. Programming sensor nodes can be a dauntingtask, especially for non-programmers. For this pur-pose, there have been several operating systemsfor programming nodes enabling the integrationof different types of sensors and catering to scala-bility, reliability and security [52].

15 http://sourceforge.net/projects/uranus-project/16 http://www.cs.cornell.edu/bigreddata/cougar/index.php17 http://www.isi.edu/div7/scadds/18 http://www.healthcare.philips.com

4.5. Challenges in adopting pervasivesensing solutions in large epidemiologicalstudies

In addition to challenges relating to market uptakeand technical questions posed by the use of perva-sive sensors, their use on large-scale clinical trialsalso faces some obstacles. The first is that of regu-latory approval and ethical issues relating to largemedical trials. Questions relating to privacy, effecton other means of monitoring and ethical conse-quences need to be fully answered before a trialis given approval, which, although necessary, couldlead to delays in terms of testing out new technol-ogies. The pricing of such sensors could be a barrierfor their adoption in large trials, especially if gov-ernment agencies are providing funding for trials.The cost-efficiency of devices also needs to beinvestigated for large-scale trials with a minimisa-tion of maintenance costs in the long run.

The large amount of ambient and wearable sen-sors used in large-scale patient studies definitelyleads to vast amounts of data collected, especiallywhen studies are performed over long periods oftime. The main challenge has moved from beingthat of obtaining data to that of developing intelli-gent algorithms to perform pattern recognition,anomaly detection and behaviour monitoring withaccurate information on compliance and surround-ing context. Conventional data-mining methodsmight not be appropriate to deal with the distrib-uted, subject-specific and context-varying natureof sensor data over such expansive parameters.Thus, data-mining for sensor data has evolved asa field of research posing several interestingquestions that are yet unanswered. An importantarea of research is that of decision-making anddesigning knowledge-based systems using such vastamounts of data. Subject-specific models, learningfrom populations based on similarity and adaptivemodels that change according to patient variationsare factors that will probably be included as suchdecision-making systems evolve to accommodatepervasive sensing data.

4.6. Emerging economies and changes inhealthcare policies

The rapid growth of markets in countries like Bra-zil, China and India has lead to a change in howthese nations use and produce innovative medicaltechnologies. This growth has increased the de-mand for better quality healthcare, in terms ofmedical equipment, medications and informationsystems. Economic changes have signalled an eraof change in living conditions, as well as daily

Can pervasive sensing address current challenges in global healthcare? 11

habits, leading to more urbanisation and a generaldecrease in activity levels. Basic administrativesystems in hospitals are being replaced by clinicalinformation systems for patient records, accounts,and pharmacy/laboratory tests.

Investment in universities and medical researchprogrammes in these countries will eventually leadto new world leaders in the production of newaffordable technologies. Currently, developing na-tions are the leading markets for smaller and moreaffordable devices. The change in policies withrespect to intellectual property protection as wellas market competition will also affect how thesenations produce and use novel technologies.

4.7. Opportunities in pervasive sensing

A paradigm shift from reactive to preventativehealthcare is taking place worldwide with perva-sive sensors playing a very important role inaddressing some of the challenges posed. This arti-cle has addressed several challenges in globalhealthcare and has surveyed existing pervasivesensing systems that are being deployed to addressthese challenges. These systems offer a variety ofexciting opportunities from a healthcare perspec-tive. In this section three of these areas will behighlighted: encouraging active lifestyles, improv-ing rehabilitation and providing better neonatalcare.

Encouraging physical activity will increasinglyplay an important role in tackling increasing obes-ity worldwide. Sensors that can detect activityare currently used in sports and their integrationwith everyday objects, including mobile phonesand watches, is ongoing. Providing feedback topeople about their activity and encouraging themto be active is crucial if these sensors were to makea difference in behavioural patterns. Personalisingfeedback to match individual factors such as age,lifestyle and health status presents both a chal-lenge and an opportunity for pervasive sensing.

The integration of pervasive sensors in rehabili-tation programs is also an area that has started tosee interesting applications. Games using sensorsare widely available now and give an idea of howsimple it is to make them part of everyday life.Encouraging patients to perform specific motionsby using these sensors and giving them feedbackon ranges of motions and forces deployed couldbe an important stage in rehabilitation, which canbe administered remotely.

In addition to surveying applications and existingsensing systems, this article has also highlightedsome of the challenges that accompany the useof sensing within large-scale epidemiological

studies. These include technical challenges, suchas energy optimisation, reliability, security andnetworking, as well as challenges from a user per-spective stressing the importance of miniaturisa-tion, ease of use and reasonable pricing. Largedatabases resulting from these studies pose a setof yet unsolved challenges in terms of designingdecision-based systems based on distributed sens-ing, and developing context aware, subject specificand adaptable models.

Such large-scale studies pose a set of questionsnot only in terms of individual, but also in termsof group behaviour in relation to lifestyle, food-intake, genetic pre-disposition to disease andinteraction with others. These sensors might helpresearchers understand how human behaviour isaffected by these conditions as well as environ-mental factors. The influence of social, psycholog-ical and ethnic factors might also be betterunderstood after observing group behaviour oflarge populations. The understanding of this typeof data will require a combined effort betweenengineers, clinicians, behaviour scientists, psychol-ogists and sociology experts.

Acknowledgement

This work was supported by the EPSRC-funded projectESPRIT (EP/H009744/1).

References

[1] Li X, Gao L, Dai L, Zhang G, Wang X, Wei Z, et al.Understanding the relationship among urbanisation, cli-mate change and human health: a case study in Xiamen. IntJ Sustain Dev World Ecol 2010;17(4):304–10.

[2] Yang GZ. Body sensor networks. Springer-Verlag; 2006.[3] Jones KE, Patel NG, Levy MA, Storeygard A, Balk D,

Gittleman JL, et al. Global trends in emerging infectiousdiseases. Nature 2008;451(7181):990–3.

[4] Wilson PW, Kannel WB. Obesity, diabetes, and risk ofcardiovascular disease in the elderly. Am J Geriatr Cardiol2002;11(2):119–24.

[5] Stirling D, O�Connor J, Judith A, Bonadies J. Geriatric falls:injury severity is high and disproportionate to mechanism. JTrauma Inj Infect Crit Care 2001;50(1):116–9.

[6] Alirol E, Getaz L, Stoll B, Chappuis F, Loutan L. Urbanisa-tion and infectious diseases in a globalised world. LancetInfect Dis 2011;11(2):131–41.

[7] Allender S, Foster C, Hutchinson L, Arambepola C. Quan-tification of urbanization in relation to chronic diseases indeveloping countries: a systematic review. J Urban Health2008;85:938–51.

[8] Pope C, Thun M, Namboodiri M, Dockery D, Evans J, SpeizerF, et al. Particulate air pollution as a predictor of mortalityin a prospective study of US adults. Am J Respir Crit CareMed 1995;151(3):669–74.

[9] Borger C, Smith S, Truffer C, Keehan S, Sisko A, Poisal J,et al. Health spending projections through 2015: changeson the horizon. Health Aff 2006;25(2):w61–73.

12 L. Atallah et al.

[10] Thorpe KE, Florence CS, Howard DH, Joski P. The impact ofobesity on rising medical spending. Health Aff MillwoodSuppl Web 2004;1:283.

[11] Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJ.Global and regional burden of disease and risk factors,2001: systematic analysis of population health data. Lancet2006;367(9524):1747–57.

[12] Alemdar H, Ersoy C. Wireless sensor networks for health-care: a survey. Comput Netw 2010;54(15):2688–710.

[13] Pantelopoulos A, Bourbakis N. A survey on wearablesensor-based systems for health monitoring and progno-sis. IEEE Trans Syst Man Cybernet C Appl Rev 2010;40(1):1–12.

[14] Tallury P, Malhotra A, Byrne LM, Santra S. Nanobioimagingand sensing of infectious diseases. Adv Drug Deliv Rev2010;62(4–5):424–37.

[15] Guendelman S, Meade K, Benson M, Chen YQ, Samuels S.Improving asthma outcomes and self-management behav-iors of inner-city children: a randomized trial of the healthbuddy interactive device and an asthma diary. Arch PediatrAdolesc Med 2002;156:114–20.

[16] Troosters T, Sciurba F, Battaglia S, Langer D, Valluri SR,Martino L, et al. Physical inactivity in patients with COPD,a controlled multi-center pilot-study. Respir Med2010;104(7):1005–11.

[17] Barlow J, Singh D, Bayer S, Curry R. A systematic review ofthe benefits of home telecare for frail elderly people andthose with long-term conditions. J Telemed Telecare2007;13(4):172–9.

[18] Atallah L, Lo B, Ali R, King R, Yang GZ. Real-time activityclassification using ambient and wearable sensors. IEEETrans Inform Technol Biomed 2009;13(6):1031–9.

[19] Fraile J, Bajo J, Corchado J, Abraham A. Applying wearablesolutions in dependent environments. IEEE Trans InformTechnol Biomed 2010;14(6):1459–67.

[20] Scanaill C, Carew S, Barralon P, Noury N, Lyons D, Lyons G.A review of approaches to mobility telemonitoring of theelderly in their living environment. Ann Biomed Eng2006;34:547–63.

[21] Cook DJ, Augusto JC, Jakkula VR. Review: ambient intel-ligence. Technologies, applications, and opportunities.Pervasive Mob Comput 2009;5:277–98.

[22] Lee T, Mihailidis A. An intelligent emergency responsesystem: preliminary development and testing of automatedfall detection. J Telemed Telecare 2005;11(4):194–8.

[23] Popescu M, Li Y, Skubic M, Rantz M. An acoustic falldetector system that uses sound height information toreduce the false alarm rate. In: 30th Annual internationalconference of the IEEE. EMBS 2008, Engineering in Medicineand Biology Society; 2008. p. 4628–31.

[24] Bourke A, Lyons G. A threshold-based fall-detection algo-rithm using a bi-axial gyroscope sensor. Med Eng Phys2008;30(1):84–90.

[25] Salomon R, Luder M, Bieber G. iFall-a new embeddedsystem for the detection of unexpected falls. In: 8th IEEEinternational conference on pervasive computing andcommunications workshops (PERCOM workshops); 2010. p.286–91.

[26] King RC, Atallah L, Wong C, Miskelly F, Yang GZ. Elderly riskassessment of falls with BSN. International workshop onwearable and implantable body sensor networks; 2010. p.30–5.

[27] Oberzaucher J, Jagos H, Zdl C, Hlauschek W, Zagler W.Using a wearable insole gait analyzing system for auto-mated mobility assessment for older people. In: Miesen-

berger K, Klaus J, Zagler W, Karshmer A, editors.Computers helping people with special needs. Lecturenotes in computer science, vol. 6180. Springer Berlin/Heidelberg; 2010. p. 600–3.

[28] Gietzelt M, Nemitz G, Wolf KH, Meyer Zu, SchwabedissenH, Haux R, et al. A clinical study to assess fall risk using asingle waist accelerometer. Inform Health Social Care2009;34(4):181–8.

[29] Grismer M. Field sensor networks and automated monitor-ing of soil-water sensors. Soil Sci 1992;154(6):482–9.

[30] Lorincz K, Welsh M, Marcillo O, Johnson J, Ruiz M, Lees J.Deploying a wireless sensor network on an active volcano.In: IEEE internet computing; 2006. p. 18–25.

[31] Martinez K, Padhy P, Elsaify A, Zou G, Riddoch A, Hart J,et al. Deploying a sensor network in an extreme environ-ment. In: IEEE international conference on sensor net-works, ubiquitous, and trustworthy computing, vol. 1;2006. p. 8.

[32] Stoianov I, Nachman L, Madden S, Tokmouline T. Pipenetawireless sensor network for pipeline monitoring. In: Pro-ceedings of the 6th international conference on informa-tion processing in sensor networks, IPSN �07. New York, NY,USA: ACM; 2007. p. 264–73.

[33] Ma Y, Richards M, Ghanem M, Guo Y, Hassard J. Airpollution monitoring and mining based on sensor grid inLondon. Sensors 2008;8(6):3601–23.

[34] Bouwstra S, Chen W, Feijs L, Oetomo SB. Smart jacketdesign for neonatal monitoring with wearable sensors. In:Proceedings of the 2009 sixth international workshop onwearable and implantable body sensor networks, BSN �09.Washington, DC, USA: IEEE Computer Society; 2009. p.162–7.

[35] Ciani O, Piccini L, Parini S, Rullo A, Bagnoli F, Marti P, et al.Pervasive technology in neonatal intensive care unit: aprototype for newborns unobtrusive monitoring. In: 30thannual international conference of the IEEE. EMBS 2008,Engineering in Medicine and Biology Society; 2008. p.1292–5.

[36] Chen W, Ayoola I, Oetomo SB, Feijs L. Non-invasive bloodoxygen saturation monitoring for neonates using reflec-tance pulse oximeter. In: Proceedings of the conference ondesign, automation and test in Europe. DATE �10, EuropeanDesign and Automation Association, 3001 Leuven, Belgium;2010. p. 1530–35.

[37] Klonoff DC. Continuous glucose monitoring. Diabetes Care2005;28(5):1231–9.

[38] Chase HP, Roberts MD, Wightman C, Klingensmith G, GargSK, Van Wyhe M, et al. Use of the glucowatch biographerinchildren with type 1 diabetes. Pediatrics 2003;111(4):790–4.

[39] Deiss D, Bolinder J, Riveline JP, Battelino T, Bosi E,Tubiana-Rufi N, et al. Improved glycemic control in poorlycontrolled patients with type 1 diabetes using real-timecontinuous glucose monitoring. Diabetes Care 2006;29(12):2730–2.

[40] Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T,McDowell M. Physical activity in the United States mea-sured by accelerometer. Med Sci Sports Exerc 2008;40(1):181–8.

[41] Nader PR, Bradley RH, Houts RM, McRitchie SL, Obrien M.The moderate-to-vigorous physical activity from ages 9 to15 years. J Am Med Assoc 2008;300(3):295–305.

[42] Darkins A, Ryan P, Kobb R, Foster L, Edmonson E, WakefieldB, Lancaster A. Continuous glucose monitoring. Telemed JE Health 2008;10(14):1118–26.

Can pervasive sensing address current challenges in global healthcare? 13

[43] Kim YK, Kim KY, Lee KH, Kim SC, Kim H, Hwang SO, Cha KC.Clinical outcomes on real-time telemetry system in devel-oping emergency medical service system. Telemed J EHealth 2011;17(4):247–53.

[44] Aziz O, Atallah L, Lo B, ElHelw M, Wang L, Yang G, et al. Apervasive body sensor network for measuring postoperativerecovery at home. Surg Innov 2007;14(2):83–90.

[45] Mihailidis A, Barbenel JC, Fernie G. The efficacy of anintelligent cognitive orthosis to facilitate handwashing bypersons with moderate to severe dementia. NeuropsycholRehabil 2004;14(1–2):135–71.

[46] Inouez Y, Kimura T, Fujita S, Noro H, Nishikawa K, YumibaT, et al. A new parameter for assessing postoperativerecovery of physical activity using an accelerometer. SurgToday 2003;33:645–50.

[47] Yoshida T, Mizuno F, Hayasaka T, Tsubota K, Wada S,Yamaguchi T. Gait analysis for detecting a leg accidentwith an accelerometer. In: Distributed diagnosis and homehealthcare, 2006. D2H2. First transdisciplinary conference;2006. p. 43–6.

[48] Skelly M, Chizeck H. Real-time gait event detection forparaplegic walking. IEEE Trans Neural Syst Rehabil Eng2001;9(1):59–68.

[49] Tong K, Granat MH. A practical gait analysis system usinggyroscopes. Med Eng Phys 1999;21(2):87–94.

[50] Bohn J, Gartner F, Vogt H. Dependability issues of pervasivecomputing in a healthcare environment. In: Proceedings ofthe first international conference on security in pervasivecomputing. Lecture notes in computer science, vol. 2082.Springer-Verlag; 2003. p. 53–70.

[51] Corchado J, Bajo J, Tapia D, Abraham A. Using heteroge-neous wireless sensor networks in a tele-monitoring systemfor healthcare. IEEE Trans Inform Technol Biomed2010;14(2):234–40.

[52] Ellul J, Lo B, Yang GZ. The BSNOS platform: a body sensornetworks targeted operating system and toolset. In: Thefifth international conference on sensor technologies andapplications (SENSORCOMM 2011); 2011. p. 381–6.