Research On Multi-wavelength Infrared Spectrum Technology For Noninvasive Blood Glucose Testing

Diabetes mellitus is a metabolic disease characterized by chronic hyperglycemia.Currently,the number of diabetics worldwide is increasing rapidly.However,there are various complications with diabetics,which are always threatening the lives of diabetic patients.The pathogenesis of diabetes is still unknown and diabetes can not be cured completely,blood glucose monitoring for diabetic patients is particularly important.As the main method of blood glucose concentration testing,biochemical testing needs to collect blood samples frequently,not only brings diabetic patients physiological pain,but also increases their risk of infection.In this paper,based on a large number of literature research in the early stage,a multi-wavelength infrared spectrum system for non-invasive blood glucose testing has been developed.Four different bands of light-emitting diodes have been selected as the light source of the system and all controlled by programmable power supply,that is,they work in turn at a certain time interval;coupling modules of the system have been also composed of quartz plane-convex mirrors and fiber coupled mirrors in order to couple radiation source of light-emitting diodes into a four-in-one optical fiber,and the light emitted by the optical fiber has penetrated earlobe of the subject and has been sensed by the photodiode at the other side;pre-amplification circuit,low-pass filter circuit and the phase-locked amplifier circuit have been introduced to the system to detect weakly signal of transmitted light of the earlobe.At last,the spectral absorbance at different bands has been obtained.The Oral Glucose Tolerance Testing(OGTT)has been also introduced to this system to build a calibration model: first,blood glucose concentration based on traditional biochemical testing and the spectral absorbance value at earlobe of a healthy subject in the fasting state have been tested;then,blood glucose concentration and the spectral absorbance value at earlobe of the subject have been also separately tested at half an hour,one hour and two hours after prepared sugar water has been drunk in one gulp by the subject;finally,the sample data,collected in the testing,has been divided into a training set and a testing set.The neural network algorithm has been introduced to build a mathematical model between the spectral absorbance and blood glucose concentration value for the training set.The sample data of the testing set has been used to evaluate the prediction performance of the built model.The results show that the multi-wavelength infrared spectrum system for non-invasive blood glucose testing developed in this paper can preliminarily predict the change trend of blood glucose concentration of the subject.The determination coefficient between the real blood glucose concentration value and the predicted value of this system developed in this paper can reach 0.813,which lays a good foundation for the further development of a system for non-invasive blood glucose testing based on near-infrared spectroscopy analysis for clinical application in the future.