血糖浓度无创监测的建模与预测  

Model and forecast of non-invasive monitoring blood glucose concentration

作　　者：吴慧飞[1] 孟立凡[1] 高晓剑[1] 高建中[1] 

机构地区：[1]中北大学,山西太原030051

出　　处：《电子测试》2010年第1期16-18,67,共4页Electronic Test

摘　　要：将近红外漫反射光谱技术应用到血糖无创监测中,实验使用傅里叶近红外光谱仪及其光纤附件采集志愿者指尖的近红外漫反射光谱。并对光谱进行平滑,基线校正和一次求导预处理,采用偏最小二乘法建立校正模型。采集漫反射光谱的同时抽适量的血样在分光计上标定血糖的实际值,并对校正模型计算值和实际标定值进行比较,结果表明个体建模的相关性很好,相关系数达到0.891以上,均方差小于0.122。Non-invasive monitoring of blood glucose based on near-infrared diffuse reflectance spectroscopy, experiments use the FT-Near-Infrared Spectrometer and optical accessories to collecte the near-infrared diffuse reflectance spectroscopy of volunteers fingertips. At the same time, the calibration model was set up using PLS method with the smoothing,baseline correction and first derivatives. When the spectrum was obtained, the actual blod glucose value of every spectrum sample was demarcated using ultraviolet spectrophotometer. The corelation betwen the calibration value and true value for single volunter is better than that for the combination of volunters, the corelative coefficients are all over 0.891,RMSECs are allless than 0.122.

关 键 词：近红外光谱 无创血糖检测 PLS 光谱预处理 

分 类 号：R446.1[医药卫生—诊断学]