Near-infrared Blood Glucose Non-invasive Detection And Correction Model

Blood glucose measurement technique is one of the popular lessons in instrument and life science fields. Near Infrared Spectroscopy analytic technique(NIRS), has some quality characters such as high speed, good precision, reagent-free, non-invasive, naturally near infrared non-invasive blood glucose measurement is one of the important issues in this filed. It will make a qualitative progress in the method of diabetes measurement, and also has a practical meaning to the treatment of this disease.This paper presents a calibration model, which was used to obtain the blood glucose concentration in the blood glucose measurement technique, has been studied in this paper. According to the NIRS and multivariate calibration theory, setting up the basic model is the main object. Root mean square error of prediction (RMSEP ) was chose to estimate the model. There were lots of noises in the process of collecting spectrum signal, those noises would influence the precision of model. Wavelet transform has been studied to denoise. After denoising, the spectrum signal has been used to set model, then, the RMSEP was decreased 8.81mg/dL. Genetic algorithms is an effective method in wavelength selection applied in building multivariate calibration model, so it has been used to optimize model, the result showed that the numbers of wavelength for building model reduced by 50%, RMSEP was reduced 10.29mg/dL. Finally, using the software matlab to design the visual analytical system for building the model, It has been approved that it is a good platform for researchers to analyze data.