Study on Compensation of Blood Oxygen Sensor’s Ambient Light Based on GA-BP Neural Network

Abstract

In order to reduce the interference of ambient light on near-infrared blood oxygen sensors (used to measure oxygen saturation of human tissue or oxygen saturation of circulating blood during Extracorporeal membrane oxygenation treatment) and improve the measurement accuracy of a blood oxygen sensor, a compensation method of ambient light interference is proposed. This method does not need an ambient light sensor or an additional circuit. The interference intensity of ambient light is calculated by changing the light intensity value emitted by the LED in the sensor and combining the light intensity value received by the sensor. The Genetic Algorithms-Back Propagation(GA-BP) neural network is modeled, the detection light intensity value and the ambient light interference evaluation values are taken as the input characteristics, the blood gas analysis true value is taken as the expected output, and the network model is used for prediction. The results show that the accuracy of this method is higher than the traditional fitting results, and the error between the network prediction value and the true value of blood gas analysis is basically within 5%.