Case-Based Reasoning in the Health Sciences: Foundations and Research Directions - Computational Intelligence in Healthcare: Advanced Methodologies - page 133

DOMAIN
Code = { MeSH
Te rms }
YEAR
PURPOSE
αȓͲ–λȔ
MEMORY
FLAGS
αȓͲ–λȔ
REASONING
αȓͲ–λȔ
SYSTEM
ESIGN
αȓͲ–λȔ
LAGS
Fig. 5. The CBR in the Health Sciences tiered classification scheme
creating a new set of descriptors, it is proposed to use the MeSH descriptors,
of which there are over 24,000 that cover just about every aspect of the health
sciences. Along with the domain, another primary means of discriminating the
relevance of an article is its publication date. Since the date plays no real role in
classifying an article, the date has no field of its own, but instead is combined
with the Domain.
Purpose. The purposes, or tasks, of CBR systems have been thoroughly dis-
cussed in many articles summarizing CBR in the Health Sciences. One of the
first papers to survey the field in 2001, by Schmidt et al., used the purpose as the
primary means to subdivide the different systems [7]. In their paper, Schmidt
et al. specified four main purposes: diagnosis, classification, planning, and tu-
toring. Later, both Holt et al. [1] and Nilsson and Sollenborn [8] used the same
four descriptors. In the early years, the majority of systems were diagnostic in
nature, but in recent years more therapeutic and treatment systems have been
developed [9].
Table 1 presents examples of purpose classifications. Planning has been re-
placed here by treatment since most of the time planning refers to treatment
planning. However, planning tasks may involve not only treatment but also other
aspects such as diagnosis assessment, which often consists of a series of exams
and labs orchestrated in a plan. Planning is a classical major task performed
by AI systems. Therefore, planning can be added to the treatment choice in the
purpose dimension.