Background: Chronic disease is defined as disease that persists over a long condition which progress slowly and generally it can be controlled but not cure. There are types of chronic disease such as heart disease, stroke, cancer, chronic respiratory disease, liver, and diabetes mellitus. CBR is a case-based reasoning solves problems by using or adapting solutions to old problems. The systems of CBR have been used for diverse purposes likely classification, diagnostic, planning, and tutoring in the field of medical. However, the trend CBR in diagnose chronic disease need to be reviewed due to the reliable and accuracy system has to be evaluated with parameter performance.
Method: In doing ...view middle of the document...
Initial Case-Based Reasoning (CBR) was triggered from the work of Schank and Abelson in 1977, where The first of CBR can be traced to Yale University and the work of Schank and Abelson in 19773. In 1994, Aamodt and Plaza4 proposed a life cycle for CBR systems which is used by other CBR researchers as a framework. At the highest level of a classification and generalist, a general CBR cycle may be described by the following:
a) Retrieve the most similar and suitable case or cases
b) Reuse the information, evidence, knowledge in that case to solve the problem
c) Revise the proposed solution with learning process
d) Retain the phase of this experience appropriately to be useful for future problem solving
Watson stated that CBR to solve problems by using or adapting solutions to old problems5. It has been used for diverse purposes likely classification, diagnostic, planning, and tutoring in the field of medical6.
A medical diagnosis is a classification process. A physician has to analyze lots of factors before diagnosing the chronic disease which makes the physician’s job difficult. In recent times, machine learning and data mining techniques have been considered to design an automatic diagnosis system for diabetes7. Recently, there are many methods and algorithms used to mine biomedical datasets for hidden information, including Neural networks (NNs), Decision Trees (DT), Fuzzy Logic Systems, Naive Bayes, SVM (Support Vector Machine), logistic regression and other methods8,9,7. These algorithms decrease the time spent for processing symptoms and producing diagnoses, making them more precise at the same time. There are a great variety of methods related to the diagnosis and classification of chronic disease in the literature.
Hence, the trend CBR in diagnose chronic disease need to be reviewed due to the reliable and accuracy system has to be evaluated with parameter performance. The aim of this review is explain the state of the art of the current problem and justify the parameter performance of CBR.
2. RELATED WORK
In this section, we describe related work with CBR in diagnose disease. The related work purposes to enrich knowledge CBR method.
Subhagata10 found that CBR able to diagnose Premenstrual Syndrome (PMS) using K-Nearest Neighbour (kNN) to search k similar cases based on Euclidean distance measure. The novelty of T.A prototype software tool is design a flexible auto-set tolerance (T).
Under research forensic Intelligent Forensic Autopsy Report System (I-AuReSys), Yeow proposed11 system based on a CBR technique with a Naïve Bayes learner for feature-weights learning.
In other case, Xiong12 stated hybrid CBR genetic-based fuzzy rule learning using Iris, Wine, and Cleveland (heart disease) data sets and finally the result shown that accuracy reach to 93.25%.
Jha stated that CBR can be applied in Diabetes Detection and Care uses 4 Diabetes Support System TM (4DSS). The Intelligence System has aimed to: (a) automatically Detect problems in...