Expert System for Non-Invasive Classification of Total Cholesterol Level Using Bioelectrical Impedance

Abstract

This paper describes the development of a Rule Based Expert System (ES) to classify the Total Cholesterol (TC) level using Bioelectrical Impedance Analysis (BIA). A total of 199 voluntary subjects were recruited in the study. The BIA parameters that are statistically significant predictors are body capacitance (BC), basal metabolic rate (BMR) extracellular mass (ECM) and lean body mass (LBM). The ES was developed using Bayesian reasoning method. The developed ES is able to classify subjects’ TC level between normal ( 5.2 mmol/L). From the analysis using 40 testing data, the system total accuracy for classifying TC level at 0.6 probability cutoff prediction was only 70.0%. The sensitivity was 67%, and specificity of 74%. From the validation data, this ES system can classify 6 from 10 subjects correctly.