Information Technology Reference
In-Depth Information
•
The components of a MAS may be running in different machines, and can be
physically located in many different geographical locations. Each of the agents
may keep a part of the knowledge required to solve the problem in a distributed
manner, such as patient records held in different departments within a hospital.
These kind of systems improve the reliability and robustness in comparison with
centralised ones [48].
•
MAS offer a natural way to include legacy systems such as clinical information
systems. For instance, Choe and Yoo [9], propose a secure multi-agent architec-
ture for accessing healthcare information through the Internet from multiple het-
erogeneous repositories. Agents also enable to process the data locally and only
exchange a set of obtained results. For instance, HealthAgents [23] is a network
of different medical centres (named contributors) with their local existing data-
bases of cases and classifiers, which are used to diagnose new cases of brain
tumours.
•
One of the main properties of an intelligent agent is sociability. Agents are able to
communicate between themselves, using some kind of standard agent communica-
tion language, in order to exchange any kind of information. In that way they can
engage in complex dialogues, in which they can negotiate, coordinate their actions
and collaborate in the solution of a problem (for example, different care units of a
hospital may collaborate in the process of patient scheduling [38]).
•
When a problem is too complex to be solved by a single system, it is usual to
decompose it in sub-problems (which will probably not be totally independent of
each other). MAS inherently support techniques of cooperative problem solving,
in which a group of agents may dynamically discuss how to partition a problem,
how to distribute the different subtasks to be solved among them, how to ex-
change information to solve possible dependences between partial solutions, and
how to combine the partial results into the solution of the original problem [31,
52]. Thus, MAS can handle the complexity of solutions through decomposition,
modelling and organising the interrelationships between components.
•
Agents can also be used to provide information to health professionals and
patients. There exist information agents, which are specialised in retrieving
information from different sources, analysing and processing the obtained data,
selecting the information in which the user is especially interested, filtering re-
dundant or irrelevant information, and presenting it to the user with an interface
adapted to the user's preferences. Those agents implement in a natural way, AI
techniques for data and knowledge processing such as data mining, intelligent
data integration, natural language text processing or formal knowledge structures
such as ontologies [37, 47]. In addition, as Vieira-Marques
et al.
[51] show, a
common relational database for all the agents is often unpractical because of cost
and technical requirements, and medical institutions usually prefer to maintain
control of their own medical data.
•
Another important property of agents is their pro-activity, their ability to per-
form tasks and execute processes that may be beneficial for the user, even if
he/she has not explicitly requested those tasks to be executed. Using this prop-
erty and taking into consideration information about the user's personal profile,
they may find relevant information and show it to the user before he has to
request it.
Search WWH ::
Custom Search