Agent Technology and Healthcare: Possibilities, Challenges and Examples of Application - Computational Intelligence in Healthcare: Advanced Methodologies - page 24

• Flexibility : agents may be dynamically created or eliminated according to the

needs of the application without affecting to the execution of their environment.
Negotiation and knowledge exchange allow the optimization of shared re-
sources.

• Existence of a standard : the Foundation for Intelligent Physical Agents (FIPA,

[18]) is an IEEE Computer Society standards committee that promotes the agent-
based technology and impulses the interoperability with other technologies by
definition communication standards. They establish the rules that have to govern
the design and implementation of a MAS in order to achieve interoperability
among heterogeneous agent-based systems.

• Existence of agent methodologies : due to the potential complexity of a MAS, it

is important to use a software engineering methodology in order to be able to de-
sign an agent architecture with the appropriate agent behaviours which fit with
the real-world requirements. Nowadays there exist several agent-oriented engi-
neering methodologies such as INGENIAS [42], Gaia [55] and Prometheus [41].

• Existence of software development tools : there are many tools to implement,

execute and manage MAS that provide some facilities (graphical tools, APIs,
examples, documentation, execution environment, debugging possibilities, etc.)
[36, 45]. From all the available tools, the most well-known and widely-used are
JADE (Java Agent Development Environment, [5]), Zeus [40], and agentTool
III [20].

Thanks to those advantages, MAS may be used in domains in which classical soft-
ware approaches may be hardly applied or may introduce serious limitations.
On the one hand, problems in which the knowledge required to solve is spatially
distributed in different locations may take profit from the inherently distributed nature
of agent technology, physically deploying some knowledge oriented agents in the data
source location and performing remote request for data or operations with these data.
On the other hand, problems in which several entities, while keeping their autono-
mous behaviour, have to join their problem solving abilities to solve a complex prob-
lem by exploit MAS' modularity in order to divide the problem or replicate it.
As a results of the several advantages which agent technology brings in this kind of
systems and their inherent capability of modelling and simulating complex real-world
environments in a natural way, multi-agents systems have been considered as the
latest software engineering paradigm [4, 5, 39].
Klein et al. [33] classified multi-agent systems according to their reliability into
closed and open systems. A closed MAS contain well-described agents designed to
work together. An open context deals with unreliable infrastructures, non-compliant
agents and emergent dysfunctions. Being the healthcare domain a critical environment
which typically deals with sensitive and private data, MAS studied in this chapter are
closed in terms of reliability and include coordination and cooperation techniques such
as auctions, negotiations, and planning methods among the different involved entities.

3 Agents Applied in Healthcare

As introduced above, agents and multi-agent systems provide interesting features that
have been used in complex domains such as healthcare. In this section, we analyse the