ABSTRACT
Most of the existing near-infrared noninvasive blood glucose detection models focus on the relationship between near-infrared absorbance and blood glucose concentration, but do not consider the impact of human physiological state on blood glucose concentration. In order to improve the performance of prediction model, particle swarm optimization (PSO) algorithm was used to train the structure paramters of back propagation (BP) neural network. Moreover, systolic blood pressure, pulse rate, body temperature and 1 550 nm absorbance were introduced as input variables of blood glucose concentration prediction model, and BP neural network was used as prediction model. In order to solve the problem that traditional BP neural network is easy to fall into local optimization, a hybrid model based on PSO-BP was introduced in this paper. The results showed that the prediction effect of PSO-BP model was better than that of traditional BP neural network. The prediction root mean square error and correlation coefficient of ten-fold cross-validation were 0.95 mmol/L and 0.74, respectively. The Clarke error grid analysis results showed that the proportion of model prediction results falling into region A was 84.39%, and the proportion falling into region B was 15.61%, which met the clinical requirements. The model can quickly measure the blood glucose concentration of the subject, and has relatively high accuracy.
Subject(s)
Blood Glucose , Neural Networks, Computer , Algorithms , HumansABSTRACT
Most of the existing near-infrared noninvasive blood glucose detection models focus on the relationship between near-infrared absorbance and blood glucose concentration, but do not consider the impact of human physiological state on blood glucose concentration. In order to improve the performance of prediction model, particle swarm optimization (PSO) algorithm was used to train the structure paramters of back propagation (BP) neural network. Moreover, systolic blood pressure, pulse rate, body temperature and 1 550 nm absorbance were introduced as input variables of blood glucose concentration prediction model, and BP neural network was used as prediction model. In order to solve the problem that traditional BP neural network is easy to fall into local optimization, a hybrid model based on PSO-BP was introduced in this paper. The results showed that the prediction effect of PSO-BP model was better than that of traditional BP neural network. The prediction root mean square error and correlation coefficient of ten-fold cross-validation were 0.95 mmol/L and 0.74, respectively. The Clarke error grid analysis results showed that the proportion of model prediction results falling into region A was 84.39%, and the proportion falling into region B was 15.61%, which met the clinical requirements. The model can quickly measure the blood glucose concentration of the subject, and has relatively high accuracy.
Subject(s)
Humans , Algorithms , Blood Glucose , Neural Networks, ComputerABSTRACT
One of the main challenges in the noninvasive sensing of blood glucose by near-infrared (NIR) spectroscopy is the background variations from light source drift, sweating, and temperature change at the human-machine interface. In this paper, a differential correction method based on the spectra from the floating-reference position and measuring position is proposed to eliminate these spectral variations from background interferences. Its effectiveness was validated by in vitro and in vivo experiments in which the diffuse reflectance of intralipid solutions and human skin was collected at the source distances of 0.6 mm and 2 mm by the custom-built system with six super-luminescent emitting diodes (SLEDs) light source. The results showed that, for the in vitro experiments of intralipid solutions, the coefficients of variations of diffuse reflectance decreased by 20.5% under all the six wavelengths after differential correction. For the in vivo experiments of oral glucose tolerance tests (OGTTs), partial least squares (PLS) regression models between glucose concentrations and the diffuse reflectance from palm skin were built, and the root mean square error of cross validation (RMSECV) decreased by 38.0% on average after the differential correction. Further, the spectra of the oral water tolerance tests (OWTTs) were collected for correlation with glucose concentration in OGTTs, and their correlation coefficients (R) decreased by 35.0% on average after the differential correction. Therefore, this differential correction method based on the spectra from the floating-reference position and measuring position can weaken the influence of background variations on the NIR spectroscopy and has promising potential in in vivo detection, especially for noninvasive blood glucose measurement.
