摘要
为了提升基于能量代谢守恒方法无创检测血糖的精度,设计新型无创血糖检测探头,采集300组临床试验数据,提出决策树和偏最小二乘法方法相结合的DE-PLS算法建立预测血糖值的模型。并采用多参数相关性分析,得出代谢率M和血糖浓度真值BG显著相关,验证了能量代谢守恒法无创检测血糖理论的可行性。最终结果表示,DE-PLS算法计算出的血糖浓度与真实血糖值的相关性达到了88.5%,证实该算法的精度高于应用于该领域的其他算法。
In order to improve the accuracy of non-invasive blood glucose measurement based on the method of con-servation of energy metabolism. The sensor for non-invasive blood glucose measurement was designed,300 groupsof clinical trial data were collected. This paper put forward DE-PLS algorithmthe which combine decision tree andpartial least squares method to handle data to establish forecast model of blood glucose values. The results of multiparameter correlation analysis show that the metabolic rate was significantly associated with the true values of theblood glucose,verified the correctness of non-invasive blood glucose measurement which based on the theory of con-servation of energy metabolism. The final results indicated the correlation between the test values calculated by theDE-PLS algorithm and the true values of the blood glucose reached 88.5%. Confirmed the accuracy of this algorithmis higher than other algorithms applied in this field.
出处
《传感技术学报》
CAS
CSCD
北大核心
2016年第6期808-812,共5页
Chinese Journal of Sensors and Actuators
基金
国家863重大基金项目(2011AA040406)
国家自然科学基金项目(61271119)
国家科技支撑计划项目(2013BA103B01)
广西高等学校科学研究项目(KY2015YB096)
关键词
无创血糖检测
能量代谢守恒法
传感器设计
决策树
偏最小二乘法
noninvasive blood glucose detection
the law of conservation of energy metabolism
sensor design
the decision tree
partial least squares
