Fuzzy Logic: An Application To Detect Chronic Kidney Disease And Failure

Abstract

In the Malaysian Dialysis and Transplant Registry 24th Report 2016, it was stated that for over the past 10 years, Malaysia has seen 100% increment in the number of new dialysis patients that suffer from chronic kidney disease (CKD). One of the identified factors that has contributed to this increment is the late detection of the CKD status among the patients. In fact, in the effort to reduce CKD patients, detecting the presence of CKD at an early stage is very crucial. Accordingly, the purpose of this study is to apply the predictive model in detecting CKD using MATLAB software (fuzzy logic toolbox). There are five steps involved in developing the model. First, the variables used as input data were identified which were blood urea nitrogen test, eGFR (estimated glomerular filtration rate) and serum creatinine test. Second, fuzzification of inputs and output was applied by using min-max normalization processing. Next, inference engine was constructed followed by rule aggregation. Lastly, the status of CKD for every patient was analysed by defuzzification of the outputs from the predictive model. In this study, 70 patients’ clinical tests were used as a set of data. The result shows that 47 out of 70 patients were detected as CKD patients. As a conclusion, early detection of CKD is very important to treat the disease at an early stage. This will allow patients to take an early action and follow up treatments or consultations with nephrologist to avoid any serious complications.