Intelligent Signal Analysis Using Case-Based Reasoning for Decision Support in Stress Management - Computational Intelligence in Healthcare: Advanced Methodologies - page 156

patients with very different physical reactions to stress and relaxation make stress a
complex area to apply biofeedback. A clinician normally supervises patients in the
implementation of biofeedback in stress medicine, where the patient's individual
profile is determined based on observed features and behavioral training is given.

2.2 Case-Based Reasoning (CBR)

Learning from the past and solving new problems based on previously solved cases is
the main approach of CBR. This approach is inspired by humans and how they some-
times reason when solving problems. A case (an experience) normally contains a
problem, a diagnosis/classification, a solution and its results. For a new problem case,
a CBR system matches the problem part of the case against cases in the case library or
case base and retrieves the solutions of the most similar cases after adapting it to the
current situation.
CBR cycle. A case represents a piece of knowledge as experience and plays an im-
portant role in the reasoning process. The case comprises unique features to describe a
problem. Cases can be presented in different ways [12]. To provide the solution of a
new case, the cases can be represented using a problem and solution structure (Case
structure A, Fig. 4). For the evaluation of a current case, cases can also contain the
outcome/result (Case structure B, Fig. 4).

Problem
description

Outcome

Solution

Case structure A

Case structure B

Fig. 4. Cases can contain a problem description and solution only or may include the re-
sult/outcome as a case structure in the medical domain [23]

Prior to the case representation many CBR systems depend on feature extraction
because of the complex data format in some domains. In case retrieval, a major phase
of CBR, matching of features between two cases plays a vital role. Cases with hidden
key features may affect the retrieval performance in a CBR system. Therefore, the
extraction of potential features to represent a case is highly recommended in develop-
ing a CBR system. However, feature extraction is becoming complicated in recent
medical CBR systems due to complex data formats where data is coming from sensor
signals and images or in the form of time series or free text [72]. A critical issue is
therefore to identify relevant features to represent a case in such domains.
Aamodt and Plaza has introduced a life cycle of CBR [13] which is a four-step
model with four Re-s, as shown in Fig. 5. The four Re-s: Retrieve, Reuse, Revise and
Retain represent the key tasks to implement such a cognitive model. These steps will
now be described here focusing on issues in medical CBR systems.