Design and implementation of a virtual assistant for ... - Springer Link

originalarbeiten

Design and implementation of a virtual assistant for healthcare professionals using pervasive computing technologies1 S. I. Ahamed, M. Sharmin, S. Ahmed, M. M. Haque, A. J. Khan

With the advancement of hand-held devices, wireless and sensor network pervasive computing has achieved a perfect momentum. Formerly, a requirement existed that was a serious impediment and threat to the mobility of users – the necessary presence of a fixed wired network. This has been resolved by the recent advances in wireless and mobile technologies, particularly Bluetooth and WiFi. The advancement of available, portable, low cost mobile devices (PDAs, cell phones, etc.) has resulted in the user’s mobility at unprecedented levels. As these devices can communicate with one another, the combined capabilities can be leveraged to form a useful new set of tools. Presently, pervasive computing is being extended into the sophisticated healthcare sector with the promise of providing an easier and more efficient mode of communication between physicians and patients or between the physicians themselves. In this paper we provide the details of our application ‘Healthcare Aide’, which has been designed to provide not only more convenience for doctor – doctor, resident doctor – doctor, patient – doctor and nurse – doctor interaction but also a smooth pathway for real-time decision making. Our pervasive middleware MARKS (Middleware Adaptability for Resource Discovery, Knowledge Usability and Self-healing) provides the underlying support for this application in a completely transparent manner. In this paper, we have also presented our survey results from users’ point of view along with performance analysis. Keywords: virtual assistant; Healthcare Aide; pervasive healthcare; MARKS

Design und Einfu¨hrung eines virtuellen Assistenten fu¨r im Gesundheitswesen ta¨tige Personen durch den Einsatz von ubiquita¨ren Computertechnologien. Die technologischen Fortschritte bei Pocketcomputern, drahtlosen Netzwerken (insbesondere Bluetooth und WiFi) und Sensoren ermo¨glichen die weitgehende Unterstu¨tzung der Mobilita¨t von Benutzern – dort wo fru¨her Einschra¨nkungen aufgrund fest verlegter Netzwerkkabel bestanden. Da Pocketcomputer immer leistungsfa¨higer und preiswerter werden und auch miteinander problemlos kommunizieren ko¨nnen, verspricht die Entwicklung entsprechender Applikationen Unterstu¨tzung im Bereich des Gesundheitswesens. In diesem Beitrag pra¨sentieren die Autoren Details zur Anwendung ,,Healthcare Aide‘‘, die zur Interaktion und EchtzeitEntscheidungsfindung bei Arzt – Arzt, Patient – Arzt, Schwester – Arzt verwendet werden kann. Die Middleware MARKS stellt die technische Plattform fu¨r diese Anwendung dar. In diesem Beitrag werden daru¨ber hinaus Studienergebnisse aus der Sicht der Benutzer zusammen mit einer Performanzanalyse pra¨sentiert. Schlu¨sselwo¨rter: virtueller Assistent; Healthcare Aide; ubiquita¨re Gesundheitsfu¨rsorge; Middleware MARKS

1. Introduction Pervasive computing (Weiser, 1993) is the concept that incorporates computing in our working and living environment in such a way that the interaction between humans and computers becomes extremely natural, and the user can get many types of data in a totally transparent manner. Considering virtual reality, which builds an artificial world in the computer, at one end of the spectrum, we can put pervasive computing, which embeds computing in the real world at the other. Along with the forward march of wireless and sensor networks, the demand curve has increased for PDAs, mobiles, smart phones, and other small hand-held devices, to reach exponential growth. With that, pervasive computing is showing its potential in almost every aspect of our life including hospital, emergency and critical situations, industry, education, and hostile battle fields, to name a few. The use of this technology in the field of health has been termed pervasive health care. The goal of pervasive health care is to provide healthcare services to everyone at any time overcoming

1

112

This is an extended version of a short workshop paper on our initial work. The initial work has been sented in workshop Ubicare 06 of PerCom 06.

|

heft 4.2006

the constraints of place, time and availability of doctors, nurses and resident doctors. Healthcare professionals need information delivery tools for accessing information at the point of patient care as well as transporting the information to other relevant sites. Personal digital assistants (PDAs), or hand-held devices demonstrate great promise as point of care information devices. A 1995 study on the use of PDAs at the point of care found that hardware constraints, such as memory capability limited their usefulness (Constellation Project, 1995). Since this study was completed, over the last 10 years, hand-held computer technology has advanced rapidly, and as of 2001 and 2002, between 26 % and 50 % of physicians used PDAs (Harris, 2002). ACP-ASIM survey finds nearly half of U.S. members use hand-held computers (ACP, 2002). This use appears higher among residents, with one recent study finding that over two-thirds of family practice

Ahamed, Sheikh Iqbal, Sharmin, Moushumi, Ahmed, Shameem, Haque, Munirul M., Dept. of Math., Stat. and Computer Science, Marquette University, PO Box 1881, Milwaukee, WI 53201-1881, USA; Khan, Ahmed J., Assistant Professor of Medicine, Medical College of Wisconsin, Milwaukee, WI, USA (E-Mail: [email protected])

e&i elektrotechnik und informationstechnik

originalarbeiten S. I. Ahamed et al. Design and implementation of a virtual assistant for healthcare professionals

residencies use hand-held computers in their training programs (J. Am. Med. Inform., 2002). Results of another survey and follow-up interviews on 88 residents in seven programs showed most residents use PDAs daily, regardless of practice or whether their program encourages PDAs. Uses include commercial medical references and personal organization software, such as calendars and address books. The results of this study suggests improvement needed in (a) providing secure clinical data for the current patients of a given resident, and (b) allaying concerns of catastrophic data loss from their PDAs (e.g. by educating residents about procedures to recover information from PDA backup files) (Barrett, Strayer, Schubart, 2004). However, hand-held personal digital assistants (PDAs) have undergone continuous and substantial improvements in hardware and graphics capabilities, making them a compelling platform for novel developments in high space occupying usage including even teleradiology. The latest PDAs have processor speeds of up to 400 MHz and storage capacities of up to 80 Gbytes with memory expansion methods. A recent paper by the department of emergency medicine at University of Alberta, Canada, shows that availability of bedside computer facility helped physicians to access information at the bedside and increased the use of Clinical Practice Guidelines Decision Support Tools. The patients also appeared to accept their use of information technology to assist in decision making (Bullard et al., 2004). Inpatient healthcare delivery involves complex processes that require interdisciplinary teamwork and frequent communication among physicians, nurses, unit secretaries, and ancillary staff. Often, these interactions are not at a nursing unit, or near a phone. A study was conducted at the St. Agnes Hospital in Baltimore, MD, regarding any potential benefit of wireless communication devices. The results identified a number of significant findings that demonstrate a healthcare benefit from a quantitative and qualitative standpoint (Breslin, Greskovich, Turisco, 2004). There are a number of research projects at various universities and research institutes related to pervasive healthcare. In the CodeBlue project (www.eecs.harvard.edu) at Harvard University, the researchers tried to provide an ad-hoc wireless infrastructure as a solution to a medical emergency situation. Using low power sensors (for providing vital data) and PDAs this project will assist the team of clinicians involved in the management of the emergency to communicate and share data more efficiently and also help in decisions making. In the WWM (Wireless Wellness Monitoring) project (Saranummi et al., 2001) a behavioral feedback system was implemented using some monitors, PDAs and a home server forming a WLAN (Wireless Local Area Network). In the CASCOM project (www.ercim.org), researchers tried to provide real-time solutions in the tele-medicine sector using mobile devices. Several application projects, working to satisfy various concerns or segments of the healthcare sector, are going on in Denmark’s ‘Centre for Pervasive Health Care’ (www.pervasivehealthcare.dk). In teleradiology a new technology has emerged to interface Picture archiving and communication systems (PACS) and PDA. PDAbased teleradiology has the potential to increase the efficiency of the radiologic work flow, increasing productivity and improving communication with referring physicians and patients (Raman et al., 2004). For its new acute care hospital, the University of California at Los Angeles is evaluating innovative technology involving high-resolution flat panel display devices configured as ‘‘network appliances’’ that can be wall mounted for use in the retrieval and display of medical images and data. Physicians and healthcare providers can log on with wireless hand-held computers, which can serve as an identification device as well as a navigational tool for selecting patient records and data. These data are displayed and manipulated on the flat panel display without the need for a keyboard or mouse. A prototype was developed with commercially available image dis-

April 2006 | 123. Jahrgang

play software, which was modified to allow the remote control of software functions from a handheld device through an infrared communication port. The system also allows navigation through the patient data in a World Wide Web-based electronic patient record. This prototype illustrates the evolution of radiologic facilities toward ‘‘shareable’’ high-quality display devices that allow more convenient and cost-effective access to medical images and related data in complex clinical environments, resulting in a paradigm shift in data navigation and accessibility (Ratib et al., 2003). For physicians, wireless connected hand-held computers are gaining popularity as point of care reference tools. The convergence of hand-held computers, the Internet, and wireless networks will enable these devices to assume more essential roles as mobile transmitters and receivers of digital medical information. In addition to serving as portable medical reference sources, these devices can be Internet-enabled, allowing them to communicate over wireless wide and local area networks. With enhanced wireless connectivity, hand-held computers can be used at the point of patient care for charge capture, electronic prescribing, laboratory test ordering, laboratory result retrieval, web access, e-mail communication, and other clinical and administrative tasks. Physicians in virtually every medical specialty have begun using these devices in various ways (Goldblum, 2002). Communication capable PDA has also made healthcare very versatile and mobile. Carrying hundreds of patient files in a suitcase makes medical street outreach to the homeless clumsy and difficult. Healthcare for the Homeless–Houston (HHH) began a case study under the assumption that tracking patient information with a personal digital assistant (PDA) would greatly simplify the process. Equipping clinicians with custom-designed software loaded onto Palm V Handheld Computers (palmOne, Inc, Milpitas, CA), Healthcare for the Homeless–Houston assessed how this type of technology augmented medical care during street outreach to the homeless in a major metropolitan area. Preliminary evidence suggests that personal digital assistants free clinicians to focus on building relationships instead of recreating documentation during patient encounters (Buck, Rochon, Turley, 2005). In this background, a lot of work is going on in researching to make PDA more efficient in meeting the ever increasing need of the medical professionals. All the existing and ongoing projects try to build a bridge between service providers (doctors, nurses, or paramedics) and some kind of external events (like a patient at home, emergency situations at some place, etc). In this way, technology is used to increase the interaction between patients and doctors, assisting service providers by giving them needed information. In our approach, we have tried to use pervasive computing technologies for situations that occur at the medical center. We have incorporated this technology using PDAs for information sharing in emergency decision making situations. This will also provide a means for patients to more fully participate with their physicians in managing their healthcare. This project will also provide a smooth pathway for better continuity of care for the patient, as well as a smarter doctor–doctor interaction. These benefits will, in turn, create a safer health care delivery system with reduced chances of medical error. A more efficient learning environment for the trainee physicians or medical students will be created. Our main focus areas are the sign out, endorsement, consultation or other patient related communications that go on constantly among the physicians and other healthcare professionals such as nurses, resident doctors working in-house. Our approach tries to adopt this promising technology in such a way that it maximizes the use of limited hospital resources, while at the same time creates a faster 21st century communication portal among the in-house physicians. To provide support for our application, a middleware named MARKS (Middleware Adaptability for Resource Discovery, Knowledge

heft 4.2006

|

113


Usability and Self-healing) (Ahamed et al., 2005; Sharmin, Ahmed, Ahamed, 2006a) has been developed. Along with several core services, it ensures the services of Knowledge Usability (Ahmed, Sharmin, Ahamed, 2005b), SAFE-RD (Sharmin, Ahmed, Ahamed, 2005a), Ubicomp Assistant Service (Ahmed, Sharmin, Ahamed, 2006), and GETS Self-healing (Ahmed, Sharmin, Ahamed, 2005a). From a user’s perspective, MARKS provides these services in a ubiquitous and transparent manner. Some preliminary work has been presented in (Sharmin et al., 2006b). In this paper, we have presented the details of our application along with extensive survey results. Several scenarios are depicted in sect. 2 where our ‘Healthcare Aide’ provides the perfect solution. Section 3 describes several healthcare projects with related goals. Characteristics and challenges are discussed in sect. 4, and functional requirements of ‘Healthcare Aide’ are reviewed in sect. 5. How ‘Healthcare Aide’ adheres to the required characteristics is presented in sect. 6. Graphical representations showing the placement of MARKS and ‘Healthcare Aide’ and some screen shots of the prototype have been added as supplements. Section 7 deals with how to use ‘Healthcare Aide’. The evaluation details have been portrayed in sect. 8, which is followed by conclusions and future directions in sect. 9. 2. Motivation Scenario 1 Dr. Jones, a resident doctor in a hospital, has fourteen patients in his care and he needs to sign out detailed information about all of his patients to his on-call colleague Dr. Smith. Normally, Dr. Jones would have to use the telephone to give verbal sign-out instructions for Dr. Smith to write down. Or he would retype all the patient information from his handheld device and somehow make it available to Dr. Smith electronically. Now with ‘‘Healthcare Aide’’, Dr. Jones can simply and with little effort transmit his patients’ data from his handheld device to Dr. Smith who is working somewhere else in the medical center. While observing the patient, Dr. Jones took some notes and in the evening when he leaves, new doctors will take charge. Before signing off, Dr. Jones needs to inform the evening shift doctors of his findings. He makes his observations and comments about the patients available to the associated evening shift doctors instantly and effortlessly with the use of Healthcare Aide.

A similar scenario has also been depicted in one of our papers (Ahmed, Sharmin, Ahamed, 2006). Situations like these require the need for applications that can assist doctors in their work and ensure patients can get emergency help when needed. The fast-paced, highly mobile environment of a hospital provides a situation whereby mobile technologies and applications can be very valuable. Presently, wireless devices are employed in hospitals for purposes ranging from accessing test results to completing administrative tasks such as admitting patients (BizJournals, 2004). However, this environment also provides a situation where information security is of paramount importance due to the highly sensitive nature of the information being transmitted. For example, each morning resident doctors meet for their morning rounds to discuss the conditions of patients. This meeting could be improved by allowing medical records and notes to be transferred from one resident doctor’s device to another’s or to a community of devices. In such an application, security is extremely important. A highly secure and trusted mechanism must be employed to ensure malicious entities do not obtain exceptionally sensitive data such as test results or personal information. In Fig. 1, personal information of patient B and a test result of patient A are being transferred from one doctors’ PDA to another. This requires highly secure transmission. Information is personal and confidential in this area and demands extensive support for maintaining privacy and security. In the first scenario, information about a patient is transferred which needs a secure transmission so that no information leak occurs in any point. In the second scenario, response from authoritative doctors should be considered. The patients’ privacy and security both play a vital role here. In the third scenario, as a better alternative approach, the patient could open her PDA and check the list of doctors that ends with ‘gyn’ (a suffix which indicates they are gynecologists). Now she could broadcast a message or chat to one=all of them about her problem and instantly select a doctor and make an appointment with him=her. This will not only simplify her effort but also allow her to communicate with more doctors in a very condensed timeframe.

Scenario 2 Suppose a senior medical student was performing his duty in the Intensive Care Unit (ICU). Suddenly he noticed something unusual for a particular patient. The medical student was not familiar with the symptoms and was not sure about whom to consult or how to manage the situation. He broadcasted a message containing the symptoms so that all the available in-house residents and other senior medical students received his message. One of the doctors, Dr. Morrice, understood the situation and could respond on the fly via his PDA. The medical student had followed the recommended procedures. Within a couple of minutes Dr. Morrice reached the ICU and took charge. Within this small passage of time, medication was able to be quickly administered which may have saved the life of the patient. Scenario 3 In a code blue (www.eecs.harvard.edu) scenario, doctors are running from all over the hospital to attend a cardiac arrest patient. While the response team is on its way through the hallways or elevators of the medical center, the resident taking care of the patient sends a brief PDA broadcast containing important medical information about the patient now in cardiac arrest. In this way, the response team will have immediate knowledge of the patient’s condition and not waste valuable life-saving time trying to find the patient’s medical data after arriving at the code blue scene.

114

|

heft 4.2006

Fig. 1. Medical application for doctors 3. Related works In a workplace study at IMSS General Hospital in Ensenada, Mexico, researchers developed a context aware mobile system for the hospital environment (Rodriguez et al., 2003; Mun~oz, 2003a; Mun~oz, 2003b) using Jabber open-source IM (Instant Messaging) server and a corresponding XMPP (Extensible Messaging and Presence



Protocol). But the use of fixed infrastructure like an IM server, the database for hospital information is restricted in use just inside the boundary of the hospital. Our application, however, is running completely on an ad hoc network without using any kind of fixed infrastructure. As a result, the active region of this application has been extended beyond the sphere of a hospital building. In (Favela et al., 2004; Tentori et al., 2005a; Tentori, Favela, Gonza´lez, 2005b), privacy issues of the public displays were discussed, as well as persons involved in the day-to-day hospital work like doctors, supervisors, or medical interns. Our approach was to address the privacy of the data being shared. We adopted a specific encryption technique to protect the privacy of data. Along with that, this application provides an authentication mechanism to ensure that the sophisticated medical information is only being seen by authorized personnel. In the CodeBlue project (www.eecs.harvard.edu), (Shnayder et al., 2005; Lorincz et al., 2004), a small, wearable wireless pulse oximeter and 2-lead EKG based on the Mica2 (www.xbow.com), MicaZ, and Telos (www.moteiv.com) sensor node platforms have been developed. Also being developed is a scalable software infrastructure for wireless medical devices. However, their approach is different from ours. They are deploying the sensor network concept which makes their system heavyweight while we have tried to make this system lightweight as well as efficient from a power, monetary, and performance standpoint. In the ‘Centre for Pervasive Health Care’ in Denmark, several health care projects like the AWARE project, the ComPUTE project (www.pervasivehealthcare.dk), Medical Visualization and Simulation (MedVisSim) are using sensor networks and pervasive computing technology. In their application ‘Pervasive Computing for Acute Medicine’, they tried to build a solution for critical and emergency situations involving emergency vehicles, dispatch centres and hospitals. They are trying to communicate information and generate graphical representations of information gathered through biomedical and other sensors. Our approach tries to assist the doctors directly through information and file sharing without using any external objects, permitting a more direct, resource-efficient solution. The scientists at University Rey Juan Carlos, Madrid are working on a project named CASCOM (Context-Aware Health Care Service Co-ordination in Mobile Computing Environments) (www.ercim.org). CASCOM is being implemented with the goal of providing ‘on the fly’ assistance in medical emergency situations. Organized into three layers, CASCOM will provide generic and secure decisions in the field of tele-medicine. However, they are developing this infrastructure for mobile and fixed networks. We tried to make our system as free from any infrastructure as possible. The dependency on a fixed powerful network is eliminated in our approach. 4. Characteristics of the Healthcare Aide c1) Privacy aware seamless service capability The system can be used by anybody in any situation at any time. It will present itself as an omnipresent service to the user. At the same time it will ensure privacy of information through its security mechanism, thus reducing the probability of malicious use of the highly confidential and personal information. c2) Continuous connectivity The system has to ensure persistent communication between the connected devices. It has to guarantee proper connectivity unless any of the users disconnects his=her device manually. c3) Energy efficient Battery power consumption places a real constraint on mobile applications. ‘Healthcare Aide’ handles this constraint with maximum care that has been proved by the graph shown in Fig. 8.


c4) Zero negative impact One primary goal of any application is to ensure that it doesn’t have any negative impact on the performance of other applications running on the specific device. Our application not only ensures this but optimizes the overall performance by minimal use of battery power and memory storage. c5) Tiny memory footprint Due to the common memory constraint of the mobile devices, it is nearly impossible to synthesize an application with all possible features while still capitalizing on tiny storage capacities. This application perfectly incorporates the well-known characteristic for pervasive software: miniature form. c6) Cost effectiveness through mobility Cost efficiency is another crucial aspect of mobile applications. As our system is working on ad-hoc devices without the help of any fixed infrastructure (such as a PC, a mainframe, or a central server), the cost has been significantly minimized. c6) Concurrent applications In our application, one user can run more than one feature at the same time, thus incorporating the characteristic known as ‘parallel processing’. c7) Secure service discovery In a wireless sensor network, a device like a PDA searches for other similar devices and discovers them. This is one of the major goals in building a distributed wireless network. But this goal, known as ‘service discovery,’ is just not enough. Each PDA has to be able to discover neighbors who are ‘secured’ and be able to communicate only with these ‘trusted’ friends. c8) Secure communication As file transfer and information sharing are the two most important features of this system, this communication procedure has to be absolutely sealed from intruders. c9) Authentification There has to be some kind of authentication procedure which can identify the correct user and protect the system from being used by any intruder. 5. Functional requirements of Healthcare Aide Requirements from the resident doctor=doctor’s point of view: (1) There is a pre-specified form so that the resident doctor can easily keep all the necessary information about the patient. (2) The doctor can take any additional notes on a patient if he=she wants. (3) A doctor can easily send all the information (or some selected information) to as many people as he=she wishes. (4) Except for the designated authenticated doctors, no one can access the database containing patient data. (5) During transfer of the patients’ information, it is compulsory to communicate wirelessly in such a way that any intruder will not be able to get the information (e.g. encrypt the file). (6) Normally the doctor writes down the information on paper. The doctor may enter that data in the database later. This two step system should be merged in such an efficient way that if the doctor writes to his=her device (similar to writing on paper), it would automatically be saved in the database. (7) There should be a way to save all the sign-out transactions over the past one year or more in a stable storage device like a desktop or other server. This information can be used for future patient care needs, for research or billing purposes, or medico-legal reasons. (8) A doctor may want to chat with other doctors.

heft 4.2006

|

115


Requirements from the medical student’s point of view: (9) Medical students will be able to download files from the instructor=doctor, provided the instructor=doctor agrees to share each file. The file might be any type (docs, images, pdf, voice files, video files, etc.). (10) They will be able to share an interesting rare case with other students in the medical center. (11) Communicate with other students about any change in their lecture schedule, by broadcasting a message to all the students scattered throughout the hospital. Requirements from the nurse’s point of view: (12) Nurses will be able to download files from the doctor, provided the doctor agrees to share each file. The file might be any type (docs, images, pdf, voice files, video files, etc.). (13) They will be able to share an interesting rare case with other students, doctors and resident doctors in the medical center. (14) Communicate with other nurses about any change in their schedule, by broadcasting a message to all the nurses, doctors scattered throughout the hospital. 6. Development of Healthcare Aide using MARKS How healthcare aide adheres all the characteristics: ‘Healthcare Aide’ encapsulates all of the characteristics mentioned in sect. 5. Some of the characteristics are supported by ‘Healthcare Aide’ itself, whereas the rest are provided by the core services. ‘Healthcare Aide’ incorporates the characteristic ‘privacy aware seamless service capability’ (c1). Provided that the wireless connectivity is available, it places itself in an ‘all-time ready’ service state that is available to serve whenever the user wants. The core services of our middleware (MARKS) support the characteristics ‘continuous connectivity’ (c2), ‘concurrent applications’ (c6) and ‘secure service discovery’ (c7). In accordance with c6 someone can give the command for transferring a file while he=she is chatting. A special ‘key’ feature has been introduced to achieve ‘secure service discovery’ (c7). When a PDA will try to discover its

neighbors, it will broadcast a special packet, and this will be recognized and responded to only by those trusted devices that have that predefined ‘key’. In sect. 8.3 we provided a detailed graphical representation in order to prove that this tool is really ‘energy efficient’ (c3) and complies with the ‘zero negative impact’ (c4). In order to comply with the memory storage constraints of PDAs, ‘Healthcare Aide’ captures only a small portion of memory. This fact – ‘tiny memory footprint’ (c5) is illustrated in the following table: Table 1. Code size of Healthcare Aide

Healthcare Aide

Lines of code

Executable file size (KB)

2111

29

Our tool doesn’t use any costly and comparatively heavy fixed infrastructure, and rewards itself with cost minimization (c6). Since several devices can join and leave arbitrarily at any time, an ad-hoc wireless infrastructure is very crucial and is conveniently supported by MARKS. When we transfer a file containing secure and sophisticated information, that file is transferred in encrypted form. At the destination end, receivers only with the decryption mechanism can read the file. Again MARKS has a vital role in merging this characteristic (c8). At present we are providing a heavily used password based authentication system to ensure this characteristic (c9). But we hope to give more time and effort in the future to this field, which is discussed in sect. 8. 7. How to use Healthcare Aide The necessary hardware and software requirements for proper functionality of Healthcare Aide are as follows: (1) Hand-held devices (for example Dell Axim X50v PDA) with Windows CE operating system. (2) Wireless card (Bluetooth or IEEE 802.11b). (3) MARKS (Sharmin, Ahmed, Ahamed, 2006a). To run ‘Healthcare Aide’ the following steps should be followed: Install healthcare aide: At the beginning, the user needs to install ‘Healthcare Aide’. Like other .NET compact framework references, ‘Healthcare Aide’ will appear as another supplementary reference.

"

Establish the required wireless connection: At this step the user has to ensure his=her connectivity by setting up a wireless connection in the desired wireless network that he needs to communicate. When the wireless connection is active, the device will automatically show the wireless networks available in the vicinity. When the authentication procedure is over, the user can communicate with other devices in that network with appropriate access permission. "

Run the middleware (MARKS): Now the user needs to execute MARKS. It will perform all the necessary steps required to make the bridge between the front end (Healthcare Aide) and the back end (Windows CE).

"

Execute the application: In this conquering step, the user has to run ‘Healthcare Aide’. Now the user can avail himself=herself of all the services contained in ‘Healthcare Aide’ by choosing different options from the simple front end.

"

Fig. 2. MARKS architecture

116

|

heft 4.2006



Fig. 3. Some screenshots of Healthcare Aide application 8. Evaluation We can evaluate in the following ways:

going through a set of tasks to evaluate its understandability and learning curve.

(1) Implement a prototype. (2) Cognitive walkthrough strategy (Cognitive Walkthrough). (3) Performance measurement. 8.1 Prototype implementation We have designed and implemented the prototype of Healthcare Aide. We have used WINCE as the operating system, a Dell Axim X50v as PDA hardware platform (processor type is Intel PXA270, speed is 624 MHz, display is 3.500 Transflective TFT color, and weight is 4.8 oz), VC#.Net Compact Framework as programming language, mobile ad-hoc mode of IEEE 802.11b as underlying wireless protocol, and SQLCE for database support. The screen shots of the prototype are shown in Fig. 3. 8.2 Cognitive walkthrough strategy To get the proper assessment of our application, we followed the cognitive walkthrough strategy (100þ applications), which encompasses one or a group of evaluators to inspect a user interface by Table 2. Characteristics of cognitive walkthrough strategy Applicable stages: design, test, and deployment Personnel needed for the evaluation Users: Some resident doctors and medical professionals

Can be conducted remotely: NO


8.2.1 Characteristics We have used the following characteristics of cognitive walkthrough strategy (shown in Table 2) for Healthcare Aide: 8.2.2 Input definition Who will be the users of the system? We selected several resident doctors and medical professionals to get the firsthand feedback about this application. What tasks will be analyzed? All the major tasks like maintaining a patient database, updating or deleting any patient information if needed, file transfer among doctors, chatting among doctors or patients, etc. have been analyzed. What is the correct action sequence for each task? First of all, we have briefly explained to each user about the sequence of tasks and what they should finally get. Also, we recommended they use the ‘‘HELP’’ section when necessary. We have followed the cognitive walkthrough strategy on our first prototype of Healthcare Aide. We collected the feedback from the users (doctors, patients, and medical professions) by giving a questionnaire (it is attached in the Appendix section). The distribution of the participants is given below in Table 3: Table 3. Distribution of participants

Usability issues covered Effectiveness: YES Efficiency: YES Satisfaction: YES Can obtain quantitative data: YES

Survey statistics (participants) Men Women –

10 4

Total

14

Technical Non Technical –

7 7

Age < 30 Age 30 – 50 Age > 50

14

7 4 3 14

heft 4.2006

|

117


Modifications were made based on their responses, and we have implemented a second prototype. Healthcare Aide has also been rated from the users’ perspective. Figures 4–6 show gender wise, age group wise and technical knowledge wise respectively. Figure 7 shows the users’ overall rating. It also shows usefulness, simplicity of use, and ease with which to input data as well as navigate from one screen to another.

Fig. 4. Rating of Healthcare Aide by users (men and women)

8.3 Performance measurement Power consumption is one of the most important performance metrics of pervasive computing applications and services, and Healthcare Aide seems to be very power-conservative. It consumes the minimal amount of battery power possible. Figure 8 establishes this as a true statement. Here we have shown two cases for each PDA: the idle case (when the PDA is ON but not doing anything) and the active case (when the application is running on the PDA). Figure 9 shows a similar performance measurement but based on signal strength, another crucial performance metric.

Fig. 5. Rating of Healthcare Aide by users according to age group Fig. 8. Power consumption by all PDAs before and after running the Healthcare Aide

Fig. 6. Rating of Healthcare Aide by knowledge of users

Fig. 9. Signal strength by all PDAs before and after running the Healthcare Aide

Fig. 7. Rating of Healthcare Aide by users as a whole

118

|

heft 4.2006

9. Conclusions and future works In this paper we have presented the details of our application ‘Healthcare Aide’. It has been designed to facilitate doctor – doctor, resident doctor – doctor, patient – doctor and nurse – doctor communication for real-time decision making and provide a convenient



environment for better healthcare. In our first prototype of ‘Healthcare Aide,’ we have tried to address some frequently occurring situations between patients and physicians that are of immense importance. This tool has been designed to create a healthier environment in the hospital through the creation of a secure passage for two-way interactions and flow of information. At the same time, this tool has proved the effectiveness and utility of our developed middleware MARKS. As a result of analyzing the feedback from users, we are planning to add the following features: "

"

"

Sometimes doctors want to add a note containing his=her comments when sending a patient record. We will provide this feature in our upcoming version. The chatting and information sharing service can be performed using voice command. Currently the mode of authentication is password-based. We are extending this authentication feature by using a fingerprint and=or signature which will ensure a more robust security service.

In the implemented prototype of our application we have tried to incorporate user friendliness but yet there are some scopes of improvement. In (Holzinger, 2004a) Holzinger showed details of developing user friendly interfaces according to UCD (User-Centered Design) method through a rapid prototyping model. We are planning to make our application more user oriented through the issues discussed in (Holzinger, Errath, 2004b; Holzinger, 2005). Our vision is to provide a heterogeneous way of communication to make a bridge between the existing fixed infrastructure and wireless approach. We believe that it will become an essential aid, not only for the medical community but for society as a whole. Appendix Questionnaire for Evaluation 1. Overall, how would you rate this application? (1 ¼ very bad, 5 ¼ excellent) 2. Effectiveness of the application (1 ¼ not effective at all, 5 ¼ very useful) 3. How easy the application to use? (1 ¼ very hard, 5 ¼ very easy) 4. How easy to give the input? (1 ¼ very hard, 5 ¼ very easy) 5. How easy to navigate the interface? (1 ¼ very hard, 5 ¼ very easy) 6. How easy to establish the initial wireless connection? (1 ¼ very hard, 5 ¼ very easy) 7. How easy to get the desired output after performing an action? (1 ¼ very hard, 5 ¼ very easy) 8. How effective the interface is for the small screen? (1 ¼ not at all, 5 ¼ very effective) 9. How effective the help section? (1 ¼ not at all, 5 ¼ very effective) 10. According to you, which feature is unnecessary? 11. According to you, which feature is most useful? 12. Will you choose this application if it is free? If not, why not? 13. Can you recommend any more features that should be incorporated in this software? 14. Any final suggestions or comments for improvements.

Acknowledgement

The authors appreciate the assistance of Dr. Kaiseruzzaman Khan, Paula Stroud, and Karl Stamm for their helpful suggestions to improve this paper.


References 100þ applications for PDA, http:==www.valusoft.com=products=100appspda.html ACP-ASIM press release, American College of Physicians, 9-3-2002 (electronic citation) Ahamed, S. I., Sharmin, M., Ahmed, S., Havice, Michael, J., Anamanamuri, Suresh (2005): An assessment tool for out of class learning using pervasive computing technologies. J. Inform. 8 (5), Sep. 2005. Ahmed, S., Sharmin, M., Ahamed, S. I. (2005a): GETS (Generic, Efficient, Transparent, and Secured) Self-healing service for pervasive computing applications. Submitted. Ahmed, S., Sharmin, M., Ahamed, S. I. (2005b): Knowledge usability and its characteristics for pervasive computing. The 2005 Int. Conf. on Pervasive Systems and Computing (PSC-05) in conjunction with the 2005 Int. Multi-conference in Computer Science and Engineering, CSREA Press, Las Vegas, NV, USA, June 2005: 206–209. Ahmed, S., Sharmin, M., Ahamed, S. I. (2006): Ubicomp assistant: an omnipresent customizable service for MARKS (Middleware Adaptability for Resource Discovery, Knowledge Usability and Self-healing). accepted for SAC 06. Barrett, J. R., Strayer, S. M., Schubart, J. R. (2004): Assessing medical residents’ usage and perceived needs for personal digital assistants. Int. J. Med. Inform. 2004 Feb, 73 (1): 25–34. BizJournals (2004): Healthy investment for hospitals: wireless nets. http:==www. bizjournals.com=industries= health_care= hospitals=2004=09=20=eastbay_story7.html Breslin, S., Greskovich, W., Turisco, F. (2004): Wireless technology improves nursing workflow and communications. Comput. Inform. Nurs. 2004 Sep–Oct; 22 (5): 75–81. Buck, D. S., Rochon, D., Turley, J. P. (2005): Taking it to the streets: recording medical outreach data on personal digital assistants. Comput. Inform. Nurs. 2005 Sep–Oct; 23 (5): 250–255. Bullard, M. J., Meurer, D. P., Colman, I., Holroyd, B. R., Rowe, B. H. (2004): Supporting clinical practice at the bedside using wireless technology. Acad. Emerg. Med. 2004 Nov; 11 (11): 1186–1192. Cognitive Walkthrough Strategy, http:==www.pages.drexel.edu=zwz22=CognWalk.htm Constellation Project: experience and evaluation of personal digital assistants in the clinical environment, Proc. of the 19th Annual Symposium on Computer Applications in Medical Care, 1995: 678. Favela, J., Rodriguez, M., Preciado, A., Gonzalez, V. (2004): Integrating context-aware public displays into a mobile hospital information system. IEEE Transactions on Information Technology in Biomedicine. 8 (3), Sep. 2004. Goldblum, O. M. (2002): Semin Cutan Med Surg. 2002 Sep; 21 (3): 190–201. Harris interactive poll results, Harris Poll (2002): Physician’s use of hand-helds increases from 15 % in 1999 to 26 % in 2001. 8-24-2002 (electronic citation). Holzinger, A. (2004a): Application of rapid prototyping to the user interface development for a virtual medical campus. IEEE Software. 21 (1), January 2004, 92–99, ISSN: 0740-7459. Holzinger, A. (2005): Usability engineering for software developers. Communications of the ACM, 48 (1) (January 2005): 71–74, ISSN: 0001-0782. http:==www.eecs.harvard.edu=~mdw=proj=codeblue= http:==www.ercim.org=publication=Ercim_News=enw60=caceres.html http:==www.moteiv.com= http:==www.pervasivehealthcare.dk=projects=index.html http:==www.xbow.com=Products=Wireless_Sensor_Networks.htm Holzinger, A., Errath, M. (2004b): Designing web-applications for mobile computers: experiences with applications to medicine. In: Stephanidis, C.; Stary, Chr. (2004): User centered interaction paradigms for universal access in the information society. Lecture Notes in Computer Science, LNCS 3196, ISBN: 3-540-23375-X. Lorincz, K., Malan, David, Fulford-Jones, Thaddeus, R. F., Nawoj, Alan, Clavel, Antony, Shnayder, V., Mainland, Geoff, Moulton, Steve, Welsh, M. (2004): Sensor networks for emergency response: challenges and opportunities. In: IEEE Pervasive Computing, Special Issue on Pervasive Computing for First Response, Oct–Dec 2004. Mu~noz, M., Gonzalez, V., Rodrı´guez, M., Favela, J. (2003a): Supporting context-aware collaboration in a hospital: an ethnographic informed design. Lecture Notes in Computer Science. Volume 2806, September 2003: 330–344. Mu~noz, M. A., Rodriguez, M., Favela, J., Martinez-Garcia, A. I., Gonzalez, V. (2003b): Context-aware mobile communication in hospitals. Computer IEEE, 2003, 36 (9): 38–46. Raman, B., Raman, R., Raman, L., Beauliu, C. F. (2004): Radiographics. 2004 Jan–Feb; 24 (1): 299–310. Department of Radiology, Stanford University School of Medicine, 300 Pasteur Dr, Stanford, CA 94305-5105, USA. Ratib, O., McCoy, J. M., McGill, D. R., Li, M., Brown, A. (2003): Use of personal digital assistants for retrieval of medical images and data on high-resolution flat panel displays. Radiographics, 2003 Jan–Feb; 23 (1): 267–272. Rodriguez, M., Favela, J., Gonzalez, V., Mu~noz, M. A. (2003): Agent-based mobile collaboration and information access in a healthcare environment. In: eHealth: Application of Computing Science in Medicine and Healthcare. Rudomin, I., Vazquez, J., Diaz, J. L. (eds.). Published by the Instituto Politecnico Nacional. Mexico, 2003. ISNB 970-36-0118-9: 133–148.

heft 4.2006

|

119


Saranummi, N., Korhonen, I., van Gils, M., Kivisaari, S. (2001): Barriers limiting the diffusion of ICT for proactive and pervasive health care. In: Proc. of the IX MEDICON, Pula, Croatia, 2001. Sharmin, M., Ahmed, S., Ahamed, S. I. (2005a): SAFE-RD (Secure, Adaptive, Fault Tolerant, and Efficient Resource Discovery) in pervasive computing environments. The IEEE Int. Conf. on Information Technology (ITCC 2005), Las Vegas, NV, USA, April 2005: 271–276. Sharmin, M., Ahmed, S., Ahamed, S. I. (2006a): MARKS (Middleware Adaptability for Resource Discovery, Knowledge Usability and Self-healing) for mobile devices of pervasive computing environments. To appear in the Proc. of Third Int. Conf. on Information Technology: New Generations (ITNG 2006), April, 2006, Las Vegas, Nevada, USA. Sharmin, M., Ahmed, S., Ahamed, S. I., Haque, M., Khan, A. J. (2006b): Healthcare Aide: towards a virtual assistant for doctors using pervasive middleware. Proc. of the 1st

120

|

heft 4.2006

Workshop on Ubiquitous and Pervasive Health Care (UbiCare 2006) in conjunction with Fourth Annual IEEE Int. Conf. on Pervasive Computer and Communications (PerCom 2006), Pisa, Italy, Mar 2006: 490–495. Shnayder, V., Bor-rong, Chen, Lorincz, K., Fulford-Jones, Thaddeus, R. F., Welsh, M. (2005): Sensor networks for medical care. Harvard University Technical Report. TR-08-05, April 2005. Tentori, M., Favela, J., Gonzalez, V., Rodrı´guez, M. (2005a): Supporting quality of privacy (QoP). In: Pervasive Computing. Proc. of Encuentro Internacional de Ciencias de la Computacion (ENC) 2005. IEEE Computer Press, Puebla, Mexico. (pdf file) Tentori, M., Favela, J., Gonza´lez, V. (2005b): Designing for privacy in ubiquitous computing environments. Proc. of UCAMI ’05. Granada, Espa~na. (pdf file) J. Am. Med. Inform. Assoc. 9 (1) (2002) 80. Weiser, M. (1993): Some computer science problems in ubiquitous computing. Communications of the ACM, 36 (7), July 1993: 75–84.
