Clinical Semantics - Computational Intelligence in Healthcare: Advanced Methodologies - page 12

Fig. 4. New architecture of the joint replacement surgery management system [2]
careful planning and execution for successful completion. Without this costly process,
there is a high risk for disruptions in system use during and immediately after the
migration. Considering the overall need for architectural migrations, there is potential
for the emergence of a vicious circle due to an ever increasing number of intercon-
nected, structured clinical systems providing increasingly complex functionality. The
migration of a single system to a new architecture often creates a cascade of changes
in other interconnected systems.
Ideally, such architectural overhauls would be avoided through technologies allow-
ing for the gradual evolution of system functionality and experience.
3 Open Clinical Agents
Considering the challenges of clinical informatics outlined in the previous section and
the fundamentals of open information management presented in the introduction, we
will next look at how these help to provide possible answers to the outlined chal-
lenges. This will be done in the context of an abstract architecture for open clinical
agents building on the three fundamental observations of open information manage-
ment. We will also look a practical case study involving the management of emer-
gency department chief complaints.
3.1 Abstract Architecture
Looking at information in clinical medicine from the expressivity angle, we note that
a level of expressivity similar to natural language is required to represent the diversity
of information inherent to the domain. One reflection of this fact are the numerous
failures at fully replacing free-text, natural language compositions in electronic medi-
cal records with conventionally structured information entry and the maintenance
problem inherent to the structured approach. With this in mind, natural language ex-
pressions are set down as the fundamental form of representation. The implicated goal
is a workable approach for natural language understanding (NLU) within the con-
straints of clinical medicine.
It can be argued that understanding of natural language derives from the seamless
interplay of many components including the lexical semantics (i.e., the possible
meanings of words), structural semantics (i.e., the ways of structuring and composit-
ing words, phrases and clauses to convey different meanings) and, most importantly,