Near Infrared Spectra Data Analysis by Using Machine Learning Algorithms

We present our latest research on Near Infrared Spectra data analysis by using Machine Learning algorithms. Near Infrared Spectroscopy has long been used in chemical analysis as well as agricultural products analysis. In this paper, we used it for in-vivo human skin measurements. We have also developed corresponding Machine Learning algorithms for the purposes of classification and regression. For classification, we have been able to classify the different Near Infrared Spectra for different skin sites. For regression, we have successfully trained different regression models and predicted the blood glucose levels from in-vivo skin measurement data. With the latest Texas Instruments DLP NIRscan Nano Evaluation Module, Near Infrared Spectroscopy shows a huge potential to be developed into a low cost, portable, and yet powerful, skin measurement tool. The NIR spectroscopy could be used for non-invasively measuring the blood glucose levels, without pricking fingers.