Robust Estimation Approach for a Compact Hemoglobin Components Analyzer Based on Fuzzy Expert Systems

Abstract

The current invasive methods to determine the total hemoglobin concentration (tHb) and the four hemoglobin components fractions dissipate time and blood. Using a few drops of blood, a compact haemoglobin components analyzer can measure these physiological variables. This paper introduces a fast robust estimation approach, which considered the essential ingredient of the analyzer. It does not depend directly on the optical equations such as Lambert Beer's law. Because using these optical equations includes defining some of unknown parameters that is considered the most difficult task. As a new trend and according to the general concept of the optical equations without defining any of their unknown parameters, the proposed approach is based on a unification of two fuzzy expert systems. One provides high interpretability and the other provides high accuracy. Wang-Mendel method and a new proposed modified one were used to generate a fuzzy rule-base with the aid of more than three hundred real measurements by the compact analyzer as well as a reference laboratory. The experimental results confirmed the effectiveness and the robustness of the developed whole fuzzy expert system for analyzing the total hemoglobin concentration and it`s four components ratios over one hundred and sixty different cases. The accuracy of the proposed approach was tested by means of Root Mean Square error (RMS) and Correlation (R) values. The RMS error of the tHb concentration was 0.82 g/dl and the correlation value was 0.98, which indicates that estimated values were highly correlated to the measurements of the reference laboratory.