Mobile computing in a hospital: the WARD-IN-HAND project
M.Ancona, G.Dodero, V. Gianuzzi DISI, Universita`di Genova, Via Dodecaneso, 35, Genova, Italy ancona@disi.unige.it, dodero@disi.unige.it, gianuzzi@disi.unige.it
F.Minuto DISEM, Universita`di Genova, Viale Benedetto XV, Genova, Italy minuto@unige.it
M.Guida TXT Ingegneria informatica, Via Socrate, 41, Milano, Italy guida@txt.it
ABSTRACT
We present the guidelines of a new project whose main task is to exploit mobile computers, connected via a Wireless Networks, in Personal Health systems, computerizing patient clinical records with sophisticated solutions for taking clinical information at the point of care.
Keywords
Mobile computing, wireless LAN, electronic patient record.
1. INTRODUCTION
Most European hospitals already have wired LANs and WANs supporting the traditional interconnection of the whole health care system (administration, patients record management, reservations of tests, inventory systems and so on). Quite often, the "bedside" medical treatments are out of it. That is, checking the various treatments, prescription and delivery of chemicals, notes taken by doctors and the like still do not receive networked computing support, unless handwritten notes are later on copied to electronic supports.
As a result, it is impossible to get the information in real-time and have it available at the patient's bedside. It can be estimated that up to 50% of the time of doctors and nurses is currently spent in filing and retrieving information and in coordinating and synchronizing among themselves. The impact of such inefficient practices on the quality of services provided by hospitals is dramatic, and their cost for the community is enormous.
This situation may be improved by use of bedside mobile computers [1,4], connected to the centralized databases (that is, with the patients, administration and logistics databases) by a wireless network. Doctors and nurses would be equipped with personal handheld PCs linked to a server.
The server will collect in a repository all clinical records and all relevant information about patients, including traditional clinical records, results of tests, etc. The server will also be connected with existing hospital systems, such as systems running in the analysis labs, treatment rooms or the logistic department. The server, on the basis of collected information, could trigger related actions (e.g. drugs stock update) and update the information on the palmtop computers of doctors and nurses, automatically assigning and scheduling tasks for them, so that alarms could be raised whenever critical tasks are being delayed.
2. EXISTING SYSTEMS
Some hospitals are already taking advantage of the mobility, flexibility and speed of wireless networks. Prototype systems have been already in use, mostly developed in collaboration with computer manufacturer industries. Among others, consider the following examples.
* Apple Computer and the US Department of Defense are sponsoring a collaborative project (ProMED) to explore the use and feasibility of Personal Digital Assistant (PDA) technology in the context of military medical practice, both within Medical Treatment Facilities and in the field.
* The Liverpool Women's Hospital wireless network relies on Netwave Technologies. This solution has allowed the hospital to implement a paperless care planning system, operating at the patient's bedside.
* Other prototypes have been experienced at the Good Samaritan Hospital (Fairlawn, OH) in cooperation with Aironet, and at the North Carolina Hospital, using Symbol's wireless network. Both use handheld computers which are wirelessly connected to the network via radio. The systems support the IEEE 802.11 wireless LAN standard and operates in the 2.4 GHz band using Frequency Hopping Spread Spectrum (FHSS) modulation.
3. WARD-IN-HAND: OBJECTIVES
The authors are at present cooperating within a EU-funded research project, named WARD-IN-HAND, which involves three hospitals, companies and Universities, aiming at the exploitation of a wireless network of mobile computers within different hospital departments
The network is a wireless LAN of sophisticated pen-based industrial terminals (like Telxon PTCs or Symbol PPTs) based on FSSH Radio Frequency communication, with a communication bandwidth up to 2Mbs. They will be connected to the already existing wired LAN. They shall bring information to the point of care and represent the most advanced application in clinical records processing.
The key features of Ward-in-hand are:
* "hands-free" fault tolerance and safety. Some of the above applications use mobile laptops, and this is not comfortable because of their weight and size; on the other hand, we rely on pen based tools and, heavily, on voice recognition. Sections 4.1 and 4.2 will provide further details on this subject.
* use of widely available hardware and software, to reduce costs and be compatible with existing systems. Our project takes into account both legacy systems, and emerging standardization efforts. The system will be compatible with existing European de-facto standards, for example patients record management in XML.
* security and privacy. Only those who have the right to access information can read and update patients files: this point will be deepened in Section 5.
4. HUMAN COMPUTER INTERFACE
The problem of data entry is the most crucial one: users are normally standing, a fact that discourages the use of a keyboard. Thus, the system supports two modes of interaction: Voice-based and pen-based.
Voice-based interaction
Voice is at present the main mode of interaction between doctors and nurses, hence it is natural to use the same mode also for computer interaction. The safety of such a system is then mandatory under two aspects: a correct recognition of spoken words, and discrimination of what is said by authorized personnel only.
Voice will be therefore transformed into commands to activate options and specific functions of clinical applications. Doctors and nurses shall speak in order to prescribe treatments or tests; to express diagnostic opinions; to post requests for actions; to declare execution of actions and their outcome, etc.; and more generally to generate all the workflow information needed in the hospital unit. To this purpose, voice recognition software must selectively recognize in a sentence the presence of words from a pre-defined lexicon (which includes several thousands words).
Depending upon the specific context of use, mechanisms to exclude voice which may be in the environment (e.g. patients' voice; nurses' voice while doctors are prescribing treatments, etc.) have to be devised and developed. To this purpose, the system will have self-diagnostic capability, thanks to a probability-based engine which will issue a warning whenever the degree of reliability of the command recognition will be lower than an acceptance threshold.
Pen-based interaction
As a secondary mode of interaction, users may interact with the system using a pen and the touch-screen. It will also serve as a back-up mode of interaction whenever the voice-based mechanisms will recognize a command with degree of certainty lower than a given threshold and there is a risk of mistake.
The more common approaches used for writing texts on a handheld computer are:
* Using sidebar lists and pop-up selection menus. An extensive use is feasible only for short lists of words, not with large dictionaries, like medical terms or drugs
* Using hand-writing recognition software. Such systems require the users to be trained for a while on writing styles; then, they write words letter by letter: a tedious task for most medical personnel
* Using a virtual keyboard drawn on the palmtop screen. A virtual keyboard lets a user to select the letters composing a word quickly through simple key pressing actions. However, use of a full-size keyboard is space wasting, on a small palmtop screen (usually 6" or smaller).
The above problems can be avoided by combining the keyboard and the menu selection approaches. The user clicks a minimal length prefix of the desired word, sufficient to extract it from the dictionary [3]; or else, as done by [5], a small menu oriented keyboard is shown, that dynamically changes its shape to display, at any time, a reduced number of selections, including the most likely ones.
5. SECURITY
Several issues related to safety, security and confidentiality are considered in the present project. Network coverage requires hardware redundancy to increase both availability and throughput; and we have already discussed voice recognition possibilities and problems.
Identification: To reduce the number of mobile computers, and hence the costs, doctors and nurses in the same unit share the available mobile computers. Thus, every hand-held PC should be operative after an initial log-in phase, where identity of staff members is ascertained (e.g. by recognizing the voice or by smart card insertion). When the user is logged-in, he/she will belong to a pre-defined security level, and the allowed operation set will be correspondingly defined (for instance, doctors may add/remove prescriptions, nurses may not).
Logging: Once electronic notes have been transmitted to the central databases, they cannot be removed nor modified. In this way, the patient clinical record may be used to evidence both individual and hospital's responsibilities in the event of litigation.
Confidentiality: Most of the medical data, circulating in the hospital's wired and wireless network, are highly confidential, hence besides authentication at client side, data must be protected also when being transmitted. The FSHH protocol is one of the most difficult to be intercepted by itself, since data packets are randomly distributed across frequencies, in accordance with a code which differs for each transmitting unit. Thus, several mobile computers transmitting at the same time will not interfere among themselves.
Electromagnetic compatibility (EMC): A further concern is that of possible interference with hospital medical devices. Today’s devices are already quite reliable with respect to this problem: statistics reported by the US FDA say that in a one-year period ending in September 1994, only .052 of 1 percent of problem reports were EMC-related. None of the prototype experiences in hospitals ever reported of such problems specifically due to a wireless network.
6. CONCLUSIONS
Ward-in-hand is a new R&D project, aiming at increased hospitals productivity and quality of service. With respect to other medical applications with mobile computers, the strong innovative points consist in the integration of state-of-the-art technologies like voice recognition and wireless networks. The strong involvement of end-users (three different hospital units) from the very beginning is a guarantee for future product usability and diffusion.
The authors from DISI have already developed another application of mobile computers, within a research project to support field archaeology. This prototype [2] has been funded by the Italian National Research Council within the National Project for Cultural Heritage; it has already been field tested in the excavations on the Greek island of Poliochni. Such an experience had significantly different issues with respect to WARD-IN-HAND, especially from the point of view of safety and security. Archaeological data has very little concerns for safety and security; the number of involved mobile computers to be found in a site is significantly smaller, and the user interface requires input of complex drawings and digital snapshots, rather than voice commands. However the previous experience will be most valuable in the definition of network parameters and user interfaces for the present project.
7. REFERENCES