Information Technology Reference
In-Depth Information
instantiations of the system in different countries could be adapted to the methods of
work in each place, the Formal Intervention Plans used in each medical centre could
take into account only the resources they have available or the ontologies can be
adapted and extended to model new care units or new procedural or medical knowl-
edge related to the healthcare party without requiring any reprogramming effort. This
suppose a great advantage in relation with ad-hoc healthcare systems implemented
over legacy systems, which represent a rigid structured hardly adaptable in front of
runtime changes.
The use of agent technology, as had been previously argued in [11, 39], certainly
seems an interesting option to consider in this kind of distributed systems. K4Care
agents model their human counterparts in a realistic and natural way, mimicking their
real HC behaviours modelled in the knowledge structures. Agents can be dynamically
deployed and managed into the system creating, for example, an Actor Agent associ-
ated to a new patient who will immediately allow and manage the interaction of the
new user with the system in a personalized manner.
In the K4Care platform, permanent Actor Agents also implement, by relying of the
underlying data structures, the persistency necessary to maintain the execution state of
user's tasks. In this way, the inherent complexity of HC processes is transparently tack-
led by decomposing and coordinating individual agent's executions. Actor Agents'
behaviour can be also tailored according to user's preferences by locally changing some
aspects of the organizational knowledge stored in the common knowledge bases. This
adds the high degree of flexibility and adaptability required when managing heteroge-
neous user profiles from the point of view of different medical organizations.
K4Care agents incorporate behaviours to coordinate their execution and negotiate
the delegation of HC actions according to the available data. Planning and data man-
agement are the most important issues tackled by the MAS in order to ensure an effi-
cient execution of procedures and intervention plans in a distributed fashion.
The distributed nature of the system also allows to deploy Actor Agents (which
may be quite numerous) through a -secured- computer network, allowing an appro-
priate load balancing, taking profit from underused or obsolete equipment and im-
proving the efficiency and robustness of the platform. The autonomy of the agents
executing patient's intervention plans permits to enact and assign tasks avoiding a
central monitor or controller which may introduce a serious bottleneck.
Moreover, in order to provide an intuitive and remote interaction with the final
user, a web interface is provided. Due to the non-trivial communication between the
agent execution environment and the web server which process user's request, an
intermediate layer composed by temporal Gateway Agents is introduced, which main-
tains the logic -session- of the user-agent communication during the logged session
and acts as a middleware, translating requests from the web world to the FIPA-
compliant agent environment.
6 Future Lines of Work to Promote the Use of Agents
in Healthcare
The applications using agent technology in health care (especially the prominent ex-
amples mentioned in sections 3.2, 4 and 5), albeit most of them remain at the
Search WWH ::
Custom Search