VIEWS: 11 PAGES: 6 POSTED ON: 3/19/2011
Clinical Applications of Handheld Computers Michael A. Grasso George Washington University School of Medicine, Washington, DC, USA firstname.lastname@example.org Abstract As the healthcare industry continues to become more distributed, healthcare organizations are increasing their reliance on mobile links to access patient information and to update their master database at the point of care. Handheld computers have evolved into a viable platform for these systems. While initial projects have shown promise, several questions remain. This article explores the unique characteristics of handheld computers with respect to user interface design and wireless access, and introduces a prototype development effort. 1. Introduction As the healthcare industry continues to become more distributed, healthcare organizations are increasing their reliance on mobile links to access patient information and to update their master databases at the point of care. With mobile computers, clinicians can instantly update patient records at the bedside to ensure that data properly reflects the most current information. In addition, data can be validated against a centralized repository as it is being entered to help eliminate medical errors, save personnel time, and minimize the need for off- line validation. Handheld computers show promise as an important platform for the development of these new applications [1,2]. However, recent surveys report that the majority of handheld computers are used primary as reference tools and for portable computation, but are rarely used to interface with electronic medical records or with a wireless service [3,4]. This article explores the unique characteristics of handheld computers with respect to user interface design and wireless access, and introduces a prototype development effort. 2. Background The Palm OS platform redefined handheld computing by providing systems with a unique balance of features. The Pocket PC is a related device based on the Windows CE operating system. For simplicity, this paper refers to both the Palm OS and Pocket PC devices as handheld computers. These are inexpensive, lightweight, highly portable devices that are small enough to fit into a shirt pocket, and which can integrate seamlessly with other computers through a direct connection or wireless network. The typical computer has a large display, a high-speed network connection, and ample processing power. In contrast, a handheld computer has a small display, a limited network connection, and minimal processing power. Careful attention must therefore be given to software design on handheld computers in order to communicate information effectively. 2.1. User interface limitations Portable data entry is an important function of handheld computers. However, handwriting recognition is often too slow and error-prone for large data entry tasks [5,6]. Most examples in the literature report guarded, but optimistic results on pen-based input [7,8,9,10]. The efficacy of handheld computers for point-of-care data entry may be significantly limited unless pen-based input is restricted through gestures, macros, selection lists, application- specific keyboards, or speech input. The strength of handheld computers may lie more in portable information display than in data entry. Examples of this includes accessing patient care documentation , decision support [11,12], and clinical references [13,14]. A major challenge is to make efficient use of the limited amount of screen space. One approach is to minimize the use of buttons, menus, and scroll bars, which can consume a considerable amount of room on small screens. Another option is to replace these screen objects, which are traditionally used for navigation only, with objects than can be used both for navigation and to provide context. An example of this is focus-and-context visualization, which divides the screen into a set of objects, where a central object provides detail, and where other objects provide thumbnail overviews that are used for navigation [15,16]. 2.2. Haptic interfaces Haptic interfaces may provide another way to maximize screen space on handheld computers [17,18,19]. A haptic environment couples the human sense of touch with a computer, and can yield interfaces with more realistic properties. The components available for handheld computers include force resistance sensors to measure touch and pressure, accelerometers to detect rotation, and proximity range sensors. The similarities between a handheld computer and a patient's medical chart are noteworthy. Both can be picked up, carried, held in one hand, and annotated with a pen or stylus. But there are many things you can do to a medical chart that you cannot do with a handheld computer. Physicians not only write on a chart, but they might crease a page to mark important findings. They flip through pages using well-developed dexterity skills. Although not recommended, they might even turn the chart upside down and shake it to see if a missing lab slip falls out. In reality, an elaborate array of physical interactions are possible with a patient's chart, most of which are not brought to bear on handheld computers. Consider a handheld device with a haptic display that embraces all of the properties of a patient's chart. The physician would not turn the page by clicking a button or a scroll bar. Instead, the physician would turn pages by brushing a finger across the top corner of the display, similar to turning a page in a book. Another common task is to fan through a chart by bending it slightly and rapidly skimming its pages. Using tilt and pressure sensors, a handheld device might implement this by measuring the angle of the device along with finger pressure along the edge, similar to fanning through a book. 2.3. Location-dependent user interface Before entering patient data on a handheld computer, the clinician must open a specific software package, search for the correct patient, and possibly download the most recent data from a central repository. All of this takes time that normally would not occur in a paper- based system, where the clinician would simply pick up the patient’s chart that is hanging on the exam room door. The need for additional navigation on handheld computers can hinder data entry and contribute to decreased efficiency of the user interface. These experiences have led to the development of user interface principles for mobile devices that require minimal user attention and employ context awareness [20,21]. For example, when a central server determines that a clinician has entered a particular patient’s room, it can automatically instruct the handheld device to run the appropriate software, download the patient’s current medical record, and display a summary of pertinent findings. Another example is to find the closest printer when downloading lab results, find the closest workstation to display radiology images, or locate the nearest surgeon for a patient consultation. Location-dependency can also be used for passive security, for example, to prohibit access to patient data when the handheld computer is not inside a particular area of the hospital. 2.4. Wireless protocols Mobile wireless solutions would allow clinicians to instantly update patient records at the point of care to ensure that the central database properly reflects the most current information. In addition, the data can be validated against a centralized repository as it is being entered to help eliminate medical errors, save personnel time, and minimize the need for off-line data validation. Several examples of point-of-care data collection have been reported for surgery [22,23], radiology , psychiatry , clinical trials , home health visits , and electronic prescriptions . Other important areas of research for the healthcare industry include security [29,30], signal interference [31,32], mobile middleware architectures, mobile agents, and mobile transactions . A broader survey can be found elsewhere . Two important wireless standards to emerge in recent years are Wi-Fi (802.11)  and Bluetooth . Both provide high bandwidth over short distances. They operate over the 2.4 GHz unlicensed band and can connect to the existing internet infrastructure through wireless access points. The Wi-Fi specification is a wireless extension of Ethernet that supports TCP/IP and other forms of network traffic. It allows for wireless transmission speeds up to 54 megabits per second (Mbps) at distances up to 1,000 feet. The Bluetooth specification is complementary to Wi-Fi. It allows electronic devices to identify services and communicate through ad hoc networks called piconets. It operates at distances up 300 feet and at transmission speeds up to 2 Mbps. Based on initial studies, signal interference with sensitive medical equipment does not seem to be a problem [22,37]. Another option is to transmit two-way data over long distances using the cellular network. Examples of this are CDPD (cellular digital packet data)  and GPRS (general packet radio service) . These protocols allow mobile users to connect to the Internet at transmission speeds of around 100 kilobits per second (Kbps) or less. This is three orders or magnitude slower than the Wi-Fi standard. However, because they use existing cellular networks, they can be deployed over long distances, similar to a pager or cell phone. 2.5. Weakly connected environments Traditionally, clinical databases have been stored in central computing facilities using fixed computers with high bandwidth, hard-wired network connections, and powerful processors. In contrast, wireless-enabled handheld computers operate in weakly connected environments . They are subject to frequent disconnections due to power constraints of the handheld computer or limited network availability. During those times of network disconnection, the handheld computer may still be operating. The clinician may perform updates on data that resides locally on the handheld device, or queue up operations for a central server once the network is available again. This situation can result in a number of problems with respect to recoverability, consistency, and durability. For example, the updates to a patient record that are entered on a disconnected device may be lost if the handheld computer experiences a catastrophic failure, resulting in recovery problems due to this single point of failure. Patient information stored locally on a handheld computer may become out of date, resulting in consistency problems until the device can reconnect to download updates. If two people are updating the same patient record simultaneously, the changes made on one disconnected device may conflict with changes made on another, resulting in database changes that are not durable. Research into mobile transactions that attempt to guarantee durability and consistency include version vectors and anti-entropy. Version vectors store a version number with each item, which is then used to resolve updates when data is exchanged . The anti-entropy approach is based on the propagation of write operations following a set of ordering and closure constraints . With both techniques, there is still the possibility of an update conflict when the database is fragmented. Several vendors provide database tools for handheld computers, including Oracle, Sybase, and Microsoft. However, none of these can fully support real-time transactions in weakly connected environments, and thus cannot guarantee data consistency or durability. 3. Materials and methods We developed a prototype Clinical Trials Information System on an iPAQ 3800 series Pocket PC with 802.11b wireless network capabilities. We deployed the server on a Pentium- based desktop running Windows XP with a Lynksys 802.11b wireless access point. The Pocket PC software was developed with Microsoft eMbedded Visual Tools 3.0. The server software was developed with Microsoft Visual Basic 6.0 and Microsoft Access 2002. To optimize the user interface, we restricted user input to numeric data and the selection of text strings from previously-defined lists. We organized the data into a hierarchy of independent display objects, and used focus-and-context visualization to display thumbnail views of objects not currently in focus. The database scheme followed an event-oriented model, with a list of observations for each patient. Other tables in the scheme contained metadata that supported the definition of observation types and data validation rules. A more detailed description can be found elsewhere . We adopted a data sharing protocol based on optimistic fragmentation. We used read-only and exclusive locks to create database fragments for the handheld clients. Each lock had a configurable expiration time. All database fragments were monitored through shadow processes on a fixed server. Our approach optimistically assumed the database was consistent unless there was an expired lock that a client failed to release. Inconsistencies were resolved by a quorum algorithm. 4. Results We developed a system that clinicians can use to collect data, and that incorporated a number of ideas to enhance the usability of handheld computers. The prototype was evaluated by 5 participants, who were asked to collect history and physical data on simulated patients. We used an exit questionnaire to collect feedback from the participants. It included questions based on a Likert scale of 1 to 7 to gauge overall reactions to the system as accurate, dependable, efficient, intuitive, and useful. Each participant was instructed that a higher value was indicative of a more positive response, and that a value of 4 was considered neutral. We computed an acceptability index (AI) for each participant as the average ranking for each question. The average AI for all participants was 5.2, with a standard deviation of 0.25. The initial feedback was generally positive. However, there were some concerns about the amount of free-form handwriting input that might be required if enumerated lists could not be defined to anticipate all possible responses. Validation rules ensured the completeness and accuracy of the data following a simple “if-then” format based on Boolean criteria. While this format worked well for identifying missing data among sets of observations and for verifying normal ranges, the participants found it difficult to specify rules to identify trends in longitudinal data. 5. Conclusions Through the Palm OS and Pocket PC platforms, handheld computing has evolved into a viable tool for ubiquitous clinical applications that are tightly coupled to the essential processes of patient care. There remain, however, several open questions. Addressing these issues will help provide the technical advancements and empirical guidelines needed to build principled applications that respond to the information technology demands of the next generation of clinical systems. We have presented a prototype system that incorporates several design principles that were optimized for handheld computers. We plan to conduct a formal evaluation of the prototype later this year to more thoroughly assess its usability. We also plan to refine our approach to validation rules, and expand the user interface to support haptic gestures and location- depended operations. Acknowledgements Michael A. Grasso (http://home.gwu.edu/~grasso) is a medical student at the George Washington University Medical Center in Washington, DC. He also holds a PhD in computer science from the University of Maryland Baltimore County. This research was supported by grant 1R43CA84797 from the National Cancer Institute. References  Wofford MM, Secan R, Herman C, Moran WP, Wofford JL. Clinical documentation: the handheld computer as a survival tool. MD Comput 1998;15(6):352-359.  Bergeron BP. The wireless web and patient care. J Med Pract Manage 2001;17(1):39-43.  Criswell DF, Parchman ML. Handheld computer use in U.S. family practice residency programs. J Am Med Inform Assoc 2002 Jan-Feb;9(1):80-86.  Carroll A, Christakis DA. Pediatricians' use of and attitudes about personal digital assistants. Pediatrics 2004;113(2):238-242.  Lapinsky SE, Weshler J, Mehta S, Varkul M, Hallett D, Stewart TE. Handheld computers in critical care. Crit Care 2001;5(4):227-231.  Bird SB, Zarum RS, Renzi FP. Emergency medicine resident patient care documentation using a hand-held computerized device. Acad Emerg Med 2001;8(12):1200-1203.  Siddiqui KM, Butcher RJ. Nuclear cardiology patient tracking using a palm-based device: methods and clinician satisfaction. J Digit Imaging. 2002 Apr;15 Suppl 1:242-5.  Carroll AE, Saluja S, Tarczy-Hornoch P. Development of a personal digital assistant (PDA) based client/server nicu patient data and charting system. Proc AMIA Symp 2001;:100-104.  Thomas SM, Overhage JM, Warvel J, McDonald CJ. A comparison of a printed patient summary document with its electronic equivalent: early results. Proc AMIA Symp. 2001;:701-5.  Rodriguez NJ, Borges JA, Soler Y, Murillo V, Colon-Rivera CR, Sands DZ, Bourie T. PDA vs. laptop: a comparison of two versions of a nursing documentation application. Proceedings of the 16th IEEE Symposium on Computer-Based Medical Systems (CBMS 2003). 2003;:201-206.  Zupan B, Porenta A, Vidmar G, Aoki N, Bratko I, Beck JR. Decisions at hand: a decision support system on handhelds. Medinfo 2001;10(Pt 1):566-70  Howse D. InfoRetriever 3.2 for Pocket PC. J Med Libr Assoc 2002 Jan;90(1):121-122.  Tschopp M, Geissbuhler A. Use of handheld computers as bedside information providers. Medinfo 2001;10(Pt 1):764-7.  Rothschild JM, Lee TH, Bae T, Bates DW. Clinician use of a palmtop drug reference guide. J Am Med Inform Assoc 2002 May-Jun;9(3):223-229.  Bjork S, Holmquist LE, Ljungstrand P, Redstrom J. PowerView: structured access to integrated information on small screens. Proceedings of CHI, Extended Abstracts 2000;:265-266.  Taivalsaari A. The Event Horizon User Interface Model for Small Devices. Sun Microsystems, TR-99-74, 1999.  Fishkin KP, Gujar A, Harrison BL, Moran TP, Want R. Embodied user interfaces for really direct manipulation. Commun ACM 2000;43(9):74-80.  Harrison, B.L., Fishkin, K.P., Gujar, A., Mochon, C., and Want, R. Squeeze me, hold me, tilt me! An exploration of manipulative user interfaces. Proceedings of CHI 1998;:17–24.  Hinckley K, Pierce J, Sinclair M, Horvitz E. Sensing techniques for mobile interaction. Proceedings of UIST 2000;:91-100.  Pascoe J, Ryan N, Morse D. Using while moving: HCI issues in fieldwork environments. ACM Transactions on Computer-Human Interaction 2000 Sep;7(3):417-437.  Marcus A, Chen E. Designing the PDA of the future. Interactions 2002 Jan;9(1):34-44.  Fowler DL, Hogle NJ, Martini F, Roh MS. The use of a personal digital assistant for wireless entry of data into a database via the internet. Surg Endosc 2002 Jan;16(1):221-223.  Oyama L, Tannas HS, Moulton S. Desktop and mobile software development for surgical practice. J Pediatr Surg 2002 Mar;37(3):477-481. [24. Reponen J, Ilkko E, Jyrkinen L, Tervonen O, Niinimaki J, et al. Initial experience with a wireless personal digital assistant as a teleradiology terminal for reporting emergency computerized tomography scans. J Telemed Telecare 2000;6(1):45-49.  Hilty DM, Luo JS, Morache C, Marcelo DA, Nesbitt TS. Telepsychiatry: an overview for psychiatrists. CNS Drugs 2002;16(8):527-48.  Eikemeier C, Grutter R, Heitmann K. A new generation of remote data entry: using wap-phones in clinical trials. Stud Health Technol Inform 2000;77:338-342.  Brian J, Jamieson S. Post-surgical cardiac patients receive new level of care. Caring 2002 Mar;21(3):28-29.  Goldblum OM. Electronic prescribing. Dermatol Online J. 2001 Feb;7(1):1.  Borisov N, Goldberg I, Wagner D. Intercepting mobile communications. Seventh Annual International Conference on Mobile Computing and Networking 2001;:180-189.  Williams J. The IEEE 802.11b security problem. IT Professional 2001;3(6):91-96.  Golmie N, Van Dyck RE, Soltanian A. Interference of Bluetooth and IEEE 802.11. Proceedings of the 4th ACM International Workshop on Modeling, Analysis and Simulation of Wireless and Mobile Systems 2001;:11-18.  Tan KS, Hinberg I. Effects of a wireless local area network (LAN) system, a telemetry system, and electrosurgical devices on medical devices in a hospital environment. Biomed Instrum Technol 2000;34(2):115-8  Varshney U, Vetter R. Emerging mobile and wireless networks. Commun ACM 2000;43(6):73-79.  Jing J, Helel AS, Elmagarmid A. Client-server computing in mobile environments. ACM Comput Surv 1999;31(2):117-157.  Wireless Ethernet Compatibility Alliance, <http://www.wi-fi.org/>.  The Official Bluetooth Website, <http://www.bluetooth.com/>.  Bum JB, Kramer JM, Johnson KB. The Palm as a real-time wide-area data-access device. Proc AMIA Symp 2001;:52-56.  Budka KC, Jiang H, Sommars SE. Cellular digital packet data networks. <http://www.lucent.com/minds/techjournal/pdf/summer_97/paper11.pdf>  All About General Packet Radio Service, <http://www.mobilegprs.com/>.  Nixon P, Cahill V. Mobile computing: technologies for a disconnected society. Internet Computing 1998;2(1):19-20.  Parker DS, Popek GJ, Rudisin G, Stoughton A, Walker BJ, et al. Detection of mutual inconsistency in distributed systems. IEEE T Software Eng 1983;9(3):240-246.  Petersen K, Spreitzer MJ, Terry DB, Theimer MM, Demers AJ. Flexible update propagation for weakly consistent replication. Proceedings of the 16th ACM Symposium on Operating Systems Principles 1997;:288-301.  Grasso MA. Structured speech input for clinical data collection. Proceedings of the 15th IEEE Symposium on Computer Based Medical Systems (CBMS 2002). 2002;:199-204.
Pages to are hidden for
"Clinical Applications of Handheld Computers"Please download to view full document