Information Technology Reference
In-Depth Information
even though a hypothesis may be successfully verified statistically, its applica-
bility may be hindered by our inability to fully construe its complete meaning.
This meaning is defined by the complete context in which the hypothesis was
formed, which includes the data sources as well as the context of the researcher
who formed the hypothesis [62].
These considerations led Pantazi et al. to propose extending the definition of
the concept of evidence in biomedicine to align it with an intuitively appealing
direction of research: CBR [62]. From this perspective, the concept of evidence
evolves to a more complete, albeit more complex, construct which emerges nat-
urally from the attempt to understand, explain and manage unique, individual
cases. This new view of the concept of evidence is surprisingly congruent with
the currently accepted idea of evidence in forensic science [62]. Here, evidence
includes the recognition of any spatio-temporal form (i.e., pattern, regularity) in
the context of a case (e.g., a hair, a fiber, a piece of clothing, a sign of struggle,
...) which may be relevant to the solution of that case. This new view, where a
body of evidence is incremental in nature and accumulates dynamically in the
form of facts about individual cases, is in striking contrast with traditional defi-
nitions of biomedical evidence. Using case data as evidence reduces sensitivity to
the issues of recency (i.e., current evidence) and validity (i.e., best evidence) [62].
The evidence gathering mechanism enabled by CBR can lead to the design
of new research hypotheses, and engender statistical experiments aimed at inte-
grating new knowledge with theory, traditionally accomplished through the sta-
tistical approach. Therefore, CBR's evidence gathering role complements that
of the statistical approach. As a matter of fact, CBR, by enabling the scientific
study of individual and contextual data elements, fills a gap in the purpose of
statistics, which is the scientific study of data.
6 Research Directions
To ascertain research directions, the first author conducted an extensive review
and analysis of the literature in the field of CBR in the Health Sciences. Included
were 117 papers from all International and European Conferences on Case-Based
Reasoning until 2008, the first six Workshops on CBR in the Health Sciences,
the two DARPA CBR workshops of 1989 and 1991, and the survey papers on
CBR in the Health Sciences. A tiered classification scheme was developed, and
the overall architecture is shown in Figure 5. Five distinct categories (domain,
purpose, memory and case management, reasoning, and system design) are de-
fined. In addition, a research theme is selected to characterize the main research
hypothesis and findings of the paper. This work was first presented at the Sev-
enth Workshop on CBR in the Health Sciences, which was held in 2009. This
section includes excerpts from the workshop paper [64].
6.1 Classification System
Domain. The range of domains in the health sciences fields is vast and, as
a result, it was chosen as the first level of classification. However, rather than
 
Search WWH ::




Custom Search