Research on a Non-invasive Neonatal Jaundice Detector

The jaundice phenomenon caused by excessive bilirubin in newborns seriously endangers their health. But the risks of infection make the blood sampling and testing unsuitable for continuous monitoring. So it is necessary to develop a non-invasive real-time detection instrument for neonatal jaundice. The existing non-invasive jaundice detectors on the market rely on measuring the spectral changes of LED light and natural white light before and after irradiation on the skin, or comparing standard color cards with skin color sensor data to calculate the concentration of bilirubin in the body. However, the above methods have the problem of insufficient detection accuracy. This research project attempts to propose a new detection method that utilizes the fluorescence reaction phenomenon after laser irradiation on objects and the characteristics of near-infrared light energy penetrating the skin surface and entering the body. It continuously measures the fluorescence reaction data of near-infrared laser irradiation on newborn skin, and uses machine learning algorithms to model an artificial neural network to establish a correlation model between fluorescence reaction data and corresponding bilirubin concentration, To predict the concentration of bilirubin in newborns using this model. Our experimental results demonstrate that compared to existing detection methods, the accuracy of detecting bilirubin concentration using this method has been improved. This indicates that infrared laser fluorescence method for non-invasive detection of bilirubin is a promising and promising technical solution worth developing.