Experimental Research Of Human-machine Interface Methods For Non-invasive Blood Glucose Measurement By Near-Infrared Spectroscopy

In this paper,the human-machine interface methods for non-invasive blood glucose detection in near-infrared spectroscopy was studied,the source of the signal detected by the blood glucose detection system developed by the research group was analyzed,and the blood glucose signal extraction ability of the system was verified.Firstly,through the analysis of biological tissue structure and experimental research,the applicability of nine common sites in non-invasive blood glucose testing was analyzed.From the perspective of biological tissue structure and stability,the forearm and the back of the hand were considered better.In vivo experiments were conducted,and the spectral stabilization time and signal-to-noise ratio of different sites were compared.The results showed that high signal-to-noise were obtained on the extension side of forearm,the dorsal side of wrist,and the back of the hand.The spectrum stabilization time of the back of the hand,forehead,and extension side of forearm was shorter.The back of the hand and the extension side of the forearm were considered to be better measurement sites through comprehensive analysis.Further in vivo experiments were conducted to examine the sources of interference in the measurement and the stability conditions of the spectrum after the addition of fixed measures of these two sites.The results showed that the extension side of forearm was the best measurement site for near-infrared non-invasive blood glucose measurement.Secondly,in vivo experiments were performed to compare the advantages and disadvantages of two measurement methods of contact and non-contact in the nearinfrared non-invasive blood glucose detection.The main factors that influencing the spectra of the two measurement methods were analyzed,and the stable processes of the contact measurement spectra and the effect of contact pressure on the spectrum were investigated.And the stable state of non-contact measurement spectrum was studied.The results showed that it would take about 1 hour for the contact measurement spectrum to enter a stable state,in the meantime the contact pressure change should be less than 0.05 N to ensure the measurement signal-to-noise ratio to be higher than 10000:1.Compared with contact measurement,non-contact measurement spectrum could enter a stable state quickly without long warm-up time.Considering the above factors,it could be considered that non-contact measurement method was better.Finally,based on the average optical path length simulated by the Monte Carlo method,the detection precision limit of the non-invasive blood glucose detection system developed by the research group was evaluated.And under the premise of strictly controlling the measurement sites to be the same,oral water and oral glucose tolerance comparison experiment were conducted to evaluate the signal detection ability of the system.Spectral data at a single radial distance and differential spectral data of two radial distances showed that the system could detect changes in human blood glucose concentration,but the spectrum was subject to strong background interference.Through the difference processing of spectral data under different radial distance,the background interference could be effectively removed and the blood glucose signal could be extracted.