Ubiquitous healthcare monitoring system using integrated triaxial accelerometer,spo2 and location sensors
IInternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013DOI:10.5121/iju.2013.4201 1
UBIQUITOUS HEALTHCARE MONITORING SYSTEMUSING INTEGRATED TRIAXIAL ACCELEROMETER , S P O AND LOCATION SENSORS
Ogunduyile O.O ., Zuva K ., Randle O.A ., Zuva T Department of Computer Science, Tshwane University of Technology, Pretoria,South Africa Department of Computer Science, University of Botswana, Gaborone, Botswana A BSTRACT
Ubiquitous healthcare has become one of the prominent areas of research inorder to address thechallenges encountered in healthcare environment. In contribution to this area, this study developed asystem prototype that recommends diagonostic services based on physiological data collected in real timefrom a distant patient. The prototype uses WBAN body sensors to be worn by the individual and an androidsmart phone as a personal server. Physiological data is collected and uploaded to a Medical HealthServer (MHS) via GPRS/internet to be analysed. Our implemented prototype monitors the activity, locationand physiological data such as SpO and Heart Rate (HR) of the elderly and patients in rehabilitation. Theuploaded information can be accessed in real time by medical practitioners through a web application. K EYWORDS
Android smart phone, Ubiquitous Healthcare, Web application server, Wireless Body Area Networks
1. I
NTRODUCTION
The challenges in healthcare system worldwide is to provide high quality service provisioning,easy accessibility and low cost service to ever increasing population particularly the elderlysuffering from age related diseases. [1]; [2][3]. These challenges are placing a strain on theexisting healthcare systems therefore they create the necessity to develop better, smarter, costeffective and healthcare systems to provide quality healthcare services at runtime. With this alarge number of people, especially the elderly and those in rehabilitation will have easier accessto the needed healthcare resources and quality-oriented healthcare services with limited financialresources. In recent times people are more health conscious and as noted by [4] the demand forbetter healthcare services is on the rise, individuals are demanding for better healthcare servicesthat can be provided through ubiquitous healthcare systems.Healthcare services that are availed to everyone independent of time and location is known asubiquitous healthcare. Ubiquitous healthcare systems holds the potential of maintaining wellness,disease management, support for independent living, prevention and prompt treatment, along withemergency intervention anytime and anywhere as and when needed [5]. Moreover, technologiesthat provide ubiquitous healthcare services will be assimilated seamlessly in our daily lives suchthat they become invisible [6]. Ubiquitous healthcare systems use a large number of environmentsand platforms including Wireless Body Area Networks (WBANs), mobile devices and wirelessgrid/cloud/web services to make healthcare services available, observable, transparent, seamless,reliable and sustainable. Using these systems medical practitioners can remotely monitor,diagnose, access vital patient symptoms, offer advice to patients, facilitate real timecommunication with patients, give patients control over their personal data and also allow patients’ access services anywhere anytime.
Accessibility to several available services from anhealthcare provider, flexibility, security and remote health data acquisitioning, service nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 2 personalization, automatic decision making and response form the requirements for ubiquitoushealthcare systems. WBANs are characterized by the deployment of biomedical sensors aroundhuman body to proactively collect the physiological data measurements and transmit themwirelessly to the base coordinator for processing [7]. WBAN environment can continuouslyinteract with the neighbouring network nodes and can access services from the web/cloud/gridenvironment to provide new services at runtime. Providing an effective service provisioning isstill a huge challenge in a WBAN environment.The system prototype Service Oriented Wireless Body Area Networks (SOWBAN) developed inthis study is an effective service provisioning mechanism in ubiquitous healthcare environment. Itis designed to be used by patients in rehabilitation and to monitor the elderly, with the help of a 3axis accelerometer and a wireless pulse oximeter (SpO ). Body movements are monitored by an accelerometer to determine the patient’s activities i.e., running, resting, walking and dangerous activities such as falling. The SpO is used to measure blood-oxygen saturation levels (SpO ) andHeart Rate (HR) [8]. The continuous observations of physiological data has the potential togreatly improve the quality of life of patients [9]. The hardware components of the prototypecomprises of WBAN nodes (hereafter referred to as BS nodes or Body Sensor nodes),implemented on arduino fio platforms to collect physiological data and transmit it wirelessly tothe Central Intelligent Node (CIN). From the CIN, physiological data of patient is then uploadedto the Medical Health Server (MHS). Physiological data received on the MHS would beequivalent to that obtained at the medical facility, if the patient were to go there for a medicalcheck-up.The prototype SOWBAN shows a system that implements fully the concept of ubiquitoushealthcare service provisioning. The prototype architecture provides another solution to thetelemedicine technology, which is high in cost in implementation and maintenance. Throughsmart designing methods, we have integrated certain services in our prototype and make them accessible at the users’ ends (patients and medical practitioner). All events and processes within our prototype are designed as services. Such service oriented designs gives our prototype theadvantages of: interoperability, reuse, efficiency and scalability. This current study aimed todescribe our system architecture to integrate WBAN with WS for ubiquitous healthcare serviceprovisioning. Our objectives are the following:(a) To enable the flawless incorporation of wireless body sensors with web servicearchitectures.(b) To support the ubiquitous provisioning of high quality and cost-effective healthcareservices in a distributed service provisioning environment.(c) To allow for real time diagnosis of the healthcare conditions of remote patients,independent of their location.The remainder of this paper is succinctly summarized as follows. In Section 2, we discuss therelated work. In Section 3, we discuss the prototype SOWBAN system architecture looking at itsdifferent layers to feature all hardware and software components. In Section 4, we present the liveimplementation of the prototype SOWBAN architecture. Section 5, concludes the paper andhighlights its contributions as well as our future work. RELATED WORK
Most of research has been put into remote vital signals acquisition using WBAN technologies. Asurvey of WBAN applications can be found in [10]. In particular, [11] developed the OnkoNetarchitecture to support healthcare services independent of time and place using mobile computingtechnology. A wearable health systems using WBAN for patient monitoring was introduced in nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 3 three (3) layers that is; physiological sensors in first layer, personal server in second, and finallythe health care servers and related services [12] . An ubiquitous Healthcare System (UHS) wasdeveloped by [13], the system consists of vital signs devices and environment sensor devices toacquire context information to monitor and manage health status of patients anytime anywhere.The framework targeted the development of four healthcare applications including self-diagnosis,remote monitoring, exercise management and emergency services. [14], proposed WBAN tosupport medical applications and the necessity for design concepts of the hardware and networkprotocols in a multi-patient monitoring environment were highlighted.An information-based probabilistic relation model among the key indicators which sequencedtheir data gathering priority and precedence in the WBAN was constructed. They furtherconstructed a cost function over the energy expenditure involved in their data gathering, andexpressed the relationship between utility gain and energy loss as a constrained optimizationproblem. [15] gave a case study providing the fundamentals of how WBAN can be used forremote data acquisition and information fusion. Wireless devices from different technologieswere made to work together in a distributed way in a smart environment. This system uses adistributed approach to add new components in execution time. [16], proposed a SecureUbiquitous Healthcare System Architecture (SUHSA) to enable real time collection of healthcaredata from WBAN. The collected data is converted into a Clinical Document Architecture (CDA)format, digitally signed, encrypted and securely transmitted over the Internet protocol MultimediaSubsystem (IMS) and Health Level 7 (HL7) messaging standards to a central hospital for patienthealth condition to be accessed by doctors. The IMS provides internet services with Quality ofService (QoS) and it integrates different services.A comprehensive discussion of the roles of telemedicine, wireless body area networks andwireless utility grid computing technologies to address the challenges of the conventionalhealthcare system can be found in [17]. Their work integrated telemedicine technologies withMobile Dynamic Virtual Communities (MDVC) into the healthgrid for cost-effective, quality andubiquitous healthcare service provisioning. The framework allowed for remote vital signsacquisition and personalized grid services discovery through a metadata and fuzzy logic basedintelligent context engine. The gFrame is promising because of its vision to integrate MDVC intothe health grid for ubiquitous, quality and cost-effective healthcare service provisioning. Thispaper reports on life prototype implementation of SOWBAN, which is a component of gFramefor remote vital signs acquisition. The results of this study were obtained from volunteers. SOWBAN SYSTEM ARCHITECTURE
The system architecture of SOWBAN consists of the following layers: BS Arduino Layer,Central Intelligent Node Layer (CINL), and Medical Health Server (MHS) Layer which is madeup of the SOWBAN applications (the web application server) and the database as shown inFigure 1. nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 4
Figure 1. SOWBAN Architecture
The BS Arduino layer consists of the BS nodes i.e., 3 axis accelerometer and pulse oximeter,implemented on arduino fio platforms to be worn on the patien t’s body . The BS nodes gatherphysiological data from patients and transmit the data through an attached WiFly radio over a Wi-Fi wireless network to the CINL. The BS nodes have the intelligence to sense, sample, processand communicate physiological data. They also satisfy design requirements of minimum weight,greatly reduced form factor, low – power consumption through configuration (which allows forprolonged ubiquitous healthcare monitoring), seamless integration into our prototype, standardbased interface protocols and patient – specific customization. The BS nodes receiveconfiguration instructions and responds to commands from the CIN which is more superior interms of intelligence [18]. Central Intelligent Node Layer (CINL)
The CINL holds the CIN which is responsible for collecting and processing of physiological datareceived from BS nodes, providing a Graphical User Interface (GUI) for the patient to viewhis/her data in real time and uploading patient data to the MHS. It communicates with the MHSvia GPRS/Internet. The architecture uses an android smart phone with operating system (OS) 2.3vas the CIN. Android OS serves as a great choice for this architecture allowing for sophisticatedreal time data processing and increasing processing power. Integrating an android smart phonewith certain sensors e.g., an accelerometer provides a way to determine mobility and GlobalPositioning System (GPS) for location determination. This makes them more suitable for a fullyintegrated BS node ubiquitous monitoring system and is a benefit to our prototype SOWBAN.We have deployed a service on the CINL to enable configuration of the BS nodes, in terms of : node registration, node initialization, node fusion and node customisation. We have also deployedservices to enable patients upload physiological data to the MHS, download information (recommendations/prescriptions) from the MHS and view the medical practitioner’s consultationtimes or routine, if uploaded to the MHS. This will enable patients’ book medical app ointmentswhen needed. nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 5 Using a simple comparison algorithm the CIN compares changes in data received from the BSnodes to determine whether to send data to the MHS or not. Data from the BS nodes are sent onlywhen the CIN detects a change in the received data (SpO parameters, activities and location).This will help save cost for the patient by reducing the number of data transmissions made.Figure 2 shows the basic data flow algorithm of the CINL.Figure 2. CINL basic data flow algorithm Medical Health Server (MHS) Layer
The MHS stores the electronic medical records of registered users and provides various servicesto the patients, medical practitioners and healthcare providers. It is the responsibility of the MHSto authenticate users, accept health monitoring session uploads, format and insert the session datainto corresponding patient medical records, analyse the data, determine serious health situationsin order to contact emergency healthcare providers, and forward new information to the patients,such as recommendations, prescribed drugs and exercises. These operations are performed autonomously, without human’s intervention by comparing the latest patient’s physiological dataupdates with already existing ones in patient’s medical record as well as comparing recommendations and prescriptions by the patient's medical practitioner or healthcare provider.The MHS consists of the web server and application server forming a web application serverwhich can be accessed through a web browser. The application component of the web application server holds the MHS’s business processes and is responsible for communications between the
CIN, web server/web services and the database.Our developed web application server is deployed on Tomcat with Apache and MySQL is used asa database. On the application component, Add patient, registers and initializes a new patientrecord on the database. View patient, checks the existing patient medical record i.e., monitoringstatus/monitoring history, received data analysis and if a patient record no longer exists on thedatabase, delete patient. Send alert, creates an alert when data analysis results in a potentialmedical condition. With the Upload info, patient recommendations, prescriptions and medical practitioner’s consultation slots can be uploaded to the MHS. Patients can view this uploadedinformation through the download info service on the CIN over the web.Two main services offered at the web server component are enterData and collectData . Toaccess these services the medical practitioners have to login first. Using the enterData enables nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 6 data to be entered into the database while collectData helps retrieve data. These services give the medical practitioner access to the patient’s record on the M
HS over any public network providedthe correct authentication details are used. Authentication login details ensure the privacy of the patient’s medical records. Privacy involves the confidentiality of patient’s data and the assurance that no information leakage from the medical records is feasible. SOWBAN IMPLEMENTATION
The process of monitoring a person's health condition requires a large amount of information. Ourdeveloped BS nodes implemented on arduino fio platforms has the capacity to collect largeamount of physiological data and transmits these data continuously. Figure 3 shows thediagrammatical description of the data transmission process of our prototype.Figure 3. Prototype SOWBAN data transmission processThe accelerometer is worn around the patient’s waist, either using a waist belt or modifiedmobile-phone carry-case. The accelerometer gives the continuous monitoring and transmission of the patient’s angular force on the X, Y and Z axis.
These movements are translated into motionstates which corresponds to the activities of the patient with the following states existing; resting(ID - 1), walking/general movement (ID - 2), running (ID - 3) and falling (ID – . nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 7 Figure 4. Workflow process of the accelerometerHere the accelerometer is initially set in sleep state and its timer is configured to wake up at 1-secintervals collecting samples at both low measurement range (i.e., ±2g) and high sensitivity on theX, Y and Z axis. This conserves power when person is in a resting state (ID – o ) because this signifiesperson is now in an active state and the accelerometer continues collecting samples in this state.Since the normal frequency of human activities is at ranges between 1 – – . The accelerometer’s measurement ranges allows for highest sensitivity outputs on the X, Y and Z axis. If measurement shows inactivity for a period oftime (e.g., 6mins) the accelerometer returns to an inactive state and restarts the work flow processall over. This work flow process ensures automatic activity monitoring, power conservation andalertness to changes in patient activities.To carry out measurement test, our accelerometer, sampling at a frequency of 50Hz on all axes isworn around a person waist using a modified mobile-phone carry-case. Figure 5 shows the testmeasurements for the ID – – al accelerations, respectively. Due to the earth’s gravitational force, there is always an output of +g present if a person is in a vertical position and vertical measurementAcclZ is taken. During test measurements, ID –
1 and ID –
2 showed acceleration movementspatterns mostly on the X and Y axis while the Z axis showed less acceleration movement patterns.Also, measurement tests for ID – – nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 8 Figure 5. Accelerometer measurements showing (a) walking (ID –
2) and (b) running (ID – or heart activity from an ECG sensor.The wireless pulse oximeter measures blood-oxygen saturation levels (SpO ) as well as heart rate(HR) . Measurements are based on Lambert Beer’s law of spectral analysis which relates the concentration of absorbent in solution to amount of light transmitted through the solution [20].Knowing the intensity, the path length and extinction co-efficient of a substance (here,oxyhemoglobin or reduced hemoglobin) at a particular wavelength, we determine oxygensaturation by measuring the light transmitted at two different wave lengths through the fingertip.This method capitalizes on the fact that, at the red region of the light spectrum around 660nmreduced hemoglobin (Hb) has higher extinction co-efficient compared to oxyhemoglobin (HbO ).While in the near – infrared region of the light spectrum around 940nm, the extinction co-efficient of Hb is low compared to HbO With the differences in extinction co-efficient, the lightabsorbed by Hb and HbO can be determined and used to calculate a ratio R, which correlates to oxygen saturation. Using lambert’s law, we obtain an expression for the ratio for light absorbe dgiven by Equation 1. (1) I and I are the light intensity at different wave length calculated by Equation 2 and Equation 3 At wavelength λ , (2) At wavelength λ (3)WhereC o is the concentration of oxyhaemoglobin (HbO )C r is the concentration of reduced oxyhaemoglobin (Hb) ( )( )
10 1 110 2 2 loglog inin
I IR I I = ( ) o o r r C C li n
I I - + = ( ) o o r r C C lin
I I - + = nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 9 α on is the extinction co- efficient of HbO2 at wavelength λ n α rn is the extinction co- efficient of Hb at wavelength λ n From Equation 1 we show that:Oxygen saturation (SpO ) = (4)By measuring the elapsed time between peaks of the infrared light signal we get a value for heartrate using Equation 5. The infrared light signal is used to calculate for heart rate because it haslow noise and can be used in different environments [21] . Heart Rate (BPM) = (5)Our pulse oximeter test calculations report hearts rates in the range of 30 – values from 0 – /HR calculations. The pulse oximeter performsall required calculations and transmits physiological data via a WiFly radio module over the Wi-Fi network to the CIN. The WiFly is attached to the arduino fio platforms to enable radiotransmission. The WiFly can connect to any router whenever the necessary security protocols arecorrectly configured. Once connected, the module will create a server and listens on IP Address – – data because medical practitioners can only assess patient’s status correctly by the medical information they receive. To achieve this, we configured an Access point (AP) to set upa wireless connection and deployed TDMA schemes to operate on the Wi- Fi’s existing MACprotocols. Our TDMA schemes results in: a scheduled data transmission process, reduction innetwork congestion, power conservation and extended battery life of BS nodes.We developed a MoSync python application to run on the android smart phone. Python forMoSync is a very effective programming language with efficient high-level data structures thatdelivers a simple but effective approach to object-oriented scripting programming [22]. TheMoSync python application gathers data sent from the BS nodes. The CIN uses the python APIs (Application Program Interfaces) to manage the AP’s processes i.e., setting up wireless connections to both BS nodes and the MHS . On receiving data from the BS nodes, the CIN usescomparison algorithms to compare data received with the last data sent. The comparisonalgorithm is programmed to send data to the MHS only when there is a difference in data justreceived from the last data sent. The CIN also enables the patient view his/her medical status inreal time through its GUI. Figure 6 shows screenshots of the GUI interface on the CIN. ( ) ( ) o r ro r r o r o
C RC C R -=+ - - - ( )
60 sec
Period nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 10
Figure 6. Screenshots of the CI
N’s GUI showing (a) patient walking (ID –
3) on accelerometermonitoring and (b) an alert message in case of an emergency on pulse oximeter monitoringTo upload patient data to the MHS, the CIN connects with the upload info service on its GUI andsends to the web application server in TCP/HTTP over GPRS/Internet. The web applicationserver mines data received and determines health risks using logistic regression. Functionality onthe web application server is implemented in HTML/PHP. The medical practitioners can see the patient’s activities and SpO health status in real time through a web application over a webbrowser. An abnormal condition triggers the alert signal which is received by both patient andmedical practitioner and in high risk situations the GPRS on the android can be used to determine the patient’s location. Our implementation of various off the shelf components and Open Source Softwares (Python, PHP, MySQL and Apache Tomcat) helps reduces overall cost of theprototype SOWBAN.The power consumption and size for the major components in the BS Arduino Layer and CINLare listed in Table 1 to give a cost evaluation. The size of the entire BS arduino layer is limited to90 *62* 22mm and powered by a Li-Ion battery pack which is selected due to its low cost andconvenience in testing and recharging. With smart implementation power consumption is greatlyreduced in the prototype development. The CINL has a more reliable and longer lasting source ofpower, therefore will not be an issue.Table 1. Cost evaluation of the prototype SOWBAN
Component Current (mA) Size (mm)
BS nodes i.e., accelerometerand pulse oximeter 0.023 3 * 5 * 190 * 60.5 * 22Arduino Fio platform 40 27.9 * 66.0 *3.5WiFly radio transmitter 38 27*18*3.1Li-Ion battery pack _ 67.5*49.5*17The cost to build the prototype SOWBAN (excluding the MHS) is around 1450 rands, whichmakes it feasible for ubiquitously monitoring the elderly and patients in rehabilitation. Based onour implementation, we can evaluate our prototype in terms of: nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 11 (a) Safety – patient’s conditions being monitored constantly and i n real time guarantees early detection in patient’s status and treatment can be administered early enough. (b) Timeliness – with real time detection in the patient’s status i.e., especially in critical conditions where the medical practitioner is notified in immediately and can contact thehealthcare providers in closest proximity to the patient for immediate attention.(c) Effectiveness – by monitoring patient’s status and providing services ubiquitously, cost spent on healthcare is reduced, patients do not have to travel the distance to a healthfacility, wait in queues at the health facility before being attended to. Effectiveness also ensures that patient’s conditions, prescriptions and recommendations are properly documented.(d) Efficiency – this is experienced in the ability for the designed prototype to provide anoverall outcome for the patients in days of survival, years of survival, improvement instatus (especially in recovering patients ) and reduction in disability (especially in someage – related illness).
5. CONCLUSION
This paper describes the implementation of a Service Oriented Architecture (SOA) thatseamlessly combines Wireless Body Area Network (WBAN) with Web services (WS) forubiquitous healthcare service provisioning. The prototype SOWBAN developed proactivelycollects body signals of remote patients to recommend diagnostic services. This prototypeprovides continuous physiological data monitoring capabilities with minimum intervention ofmedical personnel and ubiquitous accessibility to variety of services allowing distributedhealthcare resources to be massively reused for providing cost-effective services withoutindividuals physically moving to the locations of those resources. By automating thephysiological data monitoring process the most updated information of patient will be available atall times. The SOWBAN enables the monitoring of patients physiological data in real time,promising ubiquitous, yet an affordable and effective way to provide healthcare services. Thedeveloped SOWBAN is one of the first that successfully integrates body area networks with webservices for ubiquitous healthcare service provisioning. The SOWBAN ensures interoperabilitybetween heterogeneous devices and technologies, data representations, scalability and reuse. Thecontributions of this paper are:(a) It provides the effective provisioning of services within ubiquitous healthcare system.SOWBAN achieves this by seamlessly combining the WBAN technologies i.e., sensorsystems and networks, embedded engineering etc. with web services i.e. internettechnologies, apache servers etc.(b) It allows health data monitoring, through activity information such as body positions andSpO parameters, to know a patient’s health status. (c) It provides reliable and power efficient medical data transmission within the 802.11/Wi-Fi using developed TDMA schemes.The SOWBAN also ensures privacy, integrity and authentication protocols by using passwordsand encryptions because privacy is seen as a fundamental right of humans and is a sensitive issuein healthcare applications. However, future works are required to improve the quality of serviceof body sensors wireless communication, reliability of sensor nodes, security and standardizationof interfaces. In addition, further studies of different medical conditions in clinical andambulatory settings are necessary to determine specific limitations and possible new applicationsof this technology. nternational Journal of UbiComp (IJU), Vol.4, No.2, April 2013 12 R EFERENCES [1]. Olugbara, O.O., Ojo, S.O. and Adigun, M.O. (2011). A Grid enabled framework for ubiquitoushealthcare service provisioning, Advances in Grid Computing. Zoran Constantinescu (Ed.),InTech, Croatia, ISBN: 978-953-307-301-9, 229-252.[2]. Shaikh, A., Memon, M., Memon, M. and Misbahuddin, M. (2009). The Role of Service OrientedArchitecture in Telemedicine Healthcare System. International Conference on Complex,
Intelligent and Software Intensive Systems CISIS ‘09. IEEE,
Fukuoka, Japan, 208 – – – Level Sensory Data. KSII Transactions on Internet and InformationTechnology Systems 5(11). 2016 – – – – – – th IEEE InternationalConference on Information Fusion. Salamanca, Spain, 1 – (PETRA’10), Samos Greece, 42, 1 – th International Conference on e-Health Networking, Application and Services, 19-22 June,2007, Taipei, Taiwan. 202 – – – – – rd International Conference on Pervasive Computing Technologies for Healthcare, 2009.PervasiveHealth 1 – Authors Ogunduyile O.Oluwagbenga, is a lecturer atthe Tshwane University of Technology,Pretoria, South Africa Zuva T, is a lecturer at the Tshwane Universityof Technology, Pretoria, South Africa Randle O. Abayomi, is a lecturer at theTshwane University of Technology, Pretoria,South Africa.4