Hematocrit Estimation from Transduced Current Patterns Using Single Hidden Layer Feedforward Neural Networks

The hematocrit is the most important determinant of blood viscosity and affects so much to measurements of glucose value by handheld devices. This paper presents a method to estimate hematocrit value from the transduced current curves produced by glucose-oxidase reaction in electrochemical biosensors which is used for glucose measurements. The estimation is performed by using the single- hidden layer feedforward neural networks which are trained by the extreme learning machine (ELM) algorithms with the desired output values collected from accurately measured values by a hospital analysis system. This method can obtain the acceptable result that can be used to reduce or eliminate the hematocrit dependency in measurement of glucose values from the whole blood.