Neural Network approach for rapid prediction of transcutaneous oxygen saturation

Abstract

This study presented the use of Neural Network (NN) approach in the prediction of transcutaneous oxygen saturation level, StO2. This is to overcome the limitation of using conventional signal processing approaches that are computational exhaustive. The accuracy of the NN predictive model was tested on 35 sets of new noise-corrupted Monte Carlo simulation data. This study found mean absolute error of 2.91± 2.29 % in its predictions while the statistical test revealed a strong correlation between the considered features and the predictions (ρ = 0.000). This work concluded that the proposed technique could promote further advancement in the current technology specifically in the development of portable StO2 measurement system.