Method to correct selective bias for curves of time-variation average blood glucose

To understand the variation of blood glucose for normal population, a more accurate description of average glucose level changing with different age groups is needed. This paper describes the curves of time-variation average glucose using a great mass of measured data and the method to correct selective bias based on Heckman model is proposed. It is realized that Heckman model theory could be adopted to analyze and correct the bias of glucose curves. The existence of the bias and correction effectiveness is confirmed by the analysis of the significance state ratio of selective bias (IMR) and parameter estimation. The trend of curve shapes is verified with the aid of tracking data. The average error of bias-collected curve shapes of time-variation average glucose is significantly smaller than that without correction. The result shows the effectiveness of proposed correction method.