Deploying an Agent-Based Architecture for the Management of Community Care Martin D. Beer
Richard Hill
Andrew Sixsmith
School of Computing & Management Sciences Sheffield Hallam University Sheffield, United Kingdom
School of Computing & Management Sciences Sheffield Hallam University Sheffield, United Kingdom
Department of Primary Care University of Liverpool Liverpool, United Kingdom
[email protected]
[email protected]
[email protected]
ABSTRACT This paper describes a project designed to show how a global multi-agent architecture can be used to support care in the community. Services in this area are notoriously difficult to coordinate with the direct and indirect involvement of a large number of independent, autonomous agencies, each of which has its own priorities and aspirations. Also, by the very nature of the services provided the caring community is highly mobile and distributed making the collection and dissemination of information extremely difficult. The Integrated Community Care (INCA) project has been designed to show how multi-agent technology can be used to address some of these issues.
MODEL OF CARE
Direct Care Activities
Categories and Subject Descriptors
Indirect Care Activities
Organizational Activities
Routine
Assessbbment
Marketing
Emergency
Planning
Personnel
Quality of life
Quality assurance
Financial
Management
I.2.11 [Distributed Artificial Intelligence]: Multiagent Systems Figure 1: High Level Activity Model of Care
General Terms Agents in Community Care
command and control systems. Each element of care is provided by different autonomous bodies, which maintain their own individual management information systems. The overall management of the care typically resides outside of these systems, protecting individual bodies from disclosing sensitive and irrelevant information. Key to the development of a proper care regime for any client is the Individual Care Plan that provides the link between the client’s needs and the means by which they should be met. This not only includes the Direct Care activities normally associated with community care and discussed elsewhere [1], but also the indirect and organizational activities as shown in Figure 1. These need to be reviewed and updated continuously as the client’s requirements change both by monitoring and analysis of information collected and through negotiation between the agencies.
Keywords Community Care, Collaborative Care
1.
INTRODUCTION
The delivery of community-based services presents a number of organizational, managerial and logistical problems that undermine the effectiveness and efficiency of services. While community care has typically involved limited use of information technology, recent developments have seen significant application development. An information technology within community care currently involves a number of autonomous systems; home monitoring, community alarms, care management systems and emergency systems
2.
REASONS FOR AN AGENT APPROACH
Jennings et al. [3] in reviewing the ADEPT project interpreted organizations as distributed federated structures incorporating competition and coordination and proposed that agent architectures are highly suitable for mirroring this structure. This is basically the approach taken by INCA but care has to be taken not to take this analogy too far as for
Copyright is held by the author/owner(s). AAMAS’03, July 14–18, 2003, Melbourne, Australia. ACM 1-58113-683-8/03/0007 .
932
5.
example when Symonds and Kelly [5] talk about the ’myth of the market’ since in our case:
The INCA project can now demonstrate the effective deployment of a set of agents designed to cooperate in the timely and effective delivery of community care across the Agentcities network. The cooperative activities are very different from the trading situations that have been studied extensively, and so this project gives very different insights into the design and construction of large-scale multi-agent networks. In particular we can study:
• the prime consumers (the Clients are not the actual purchasers • occupational regulation overrides market regulation • legal obligation overrides market opportunities • risk to others makes it a social issue rather than an individual matter.
• the architectural consequences of using an open multiagent system to support an application domain that requires high levels of reliability and security
All these are prime research issues for the agent community and need to be addressed if widespread deployment of any but the most basic trading agent-based solutions are to come about. In particular, the AgentCities approach can:
• the interactions between the various autonomous information sources and the architectural requirements for their reliable interoperation with the various agents.
• Integrate specific functions into a single, comprehensive, framework for service delivery. • Enhance information sharing between all parties in a controlled and open manner.
• the use of mobile and static agents to deliver the necessary services and the implications for doing this on a global platform
• Develop new services, particularly those aimed at enhancing quality of life.
• scalability issues as any realistic scenario for deploying INCA would require the linking of thousands of home units and hundreds of carers.
• Provide flexible, rather than dedicated, solutions. The objective is to provide a very much more responsive and effective service, in which valuable time and resources are not lost in duplicating actions.
3.
6.
ACKNOWLEDGEMENTS
The programming of the demonstrator has been undertaken by Iain Anderson, Wei Huang and Philip Doherty as the project work required as part of their studies. It is also in receipt of an Agentcities Deployment Grant from the European Union Agentcities.rtd project (IST-2000-28385).
USING THE DEMONSTRATOR
The current demonstrator deploys home agents across the Agentcities network. They then communicate with the other agents, located on the main platform. These are installed together with the Zeus agent platform [4]. A minimal Individual Care Plan can then be prepared. This deals solely with the handling of emergency alerts which are routed directly to the Emergency Services. The system is thus able to respond to basic requests for help immediately. Once this initial plan is entered in the database the user is able to access it through a web interface that allows it to be modified as required to meet individual need and the home unit agent can be used to assess the effectiveness of the INCA architecture to meet that need.
4.
CONCLUSIONS
7.
REFERENCES
[1] Beer, M. D, Huang W. & Sixsmith, A. ”Using Agents to Build a Practical Implementation of the INCA (Intelligent Community Alarm) System”, in L. C. Jain & Z. Chen, & N. Ichalkaranje, ”Intelligent Agents & their Applications”, Springer (2002), pp320-345. [2] Haigh, K. Z. & Yanco, H., (2002), ”Automation as Caregiver: A Survey of Issues and Technologies”, in AAAI-02 Workshop on Automation as Caregiver: The Role of Intelligent Technology in Elder Care, pages 39-53. July 2002. [3] N. R. Jennings and T. J. Norman and P. Faratin (1998), ’ADEPT: An Agent-based Approach to Business Process Management’, ACM SIGMOD Record, 27:4, 32-39. [4] Nwana, H., Ndumu, D., Lee, L., and Collis, J., (1999) ’ZEUS: A Toolkit for Building Distributed Multi-Agent Systems’, Applied Artifical Intelligence Journal, vol. 13, no. 1 pp129- 186. [5] Symonds, A. & Kelly, A. (Eds) (1998), ’The Social Construction of Community Care’, Macmillan Press Ltd., London, UK. [6] Willmott, S. et al, (2002), ’Agentcities Network Architecture’, in Proceedings of the first International Workshop on Challenges in Open Agent Systems, July 2002
FURTHER WORK
There are a number of additional agents that could beneficially be deployed to demonstrate the capabilities and flexibility of the INCA architecture. These might include: • mobile agents that emulate wireless connected Personal Digital Assistants (PDAs) • a global monitor that shows the home agents connected and their status at any time. • customized help based on information agents that know of the client’s requirements. • monitoring facilities not only for the home. • differing configurations to deal with a range of circumstances. The current demonstrator is concerned primarily with the delivery of a specified service level. It therefore deals with the communication and negotiation between the various interested parties rather than the detailed operation of the sensor networks in the home. There are other projects that are concerned more directly with this.[2]
933
