To estimate the time course of the average number of taste buds on a single FP, several regression models were constructed. A generalized linear model (GLM) with a logistic link function can be used to fit curves.14 Therefore, y = M/ [1 + exp(ax + b)], where x is postoperative months, y is the mean number of taste buds on a single FP, M is the maximal number of taste buds on “normal” FP, and parameters a and b are estimated from the GLM. The linear model, y = ax + b, with x obeys a Gaussian distribution, and the null model, y = b, were also estimated. In all models, we calculated Akaike’s Information Criterion (AIC) to clarify which model describes the observation better.15 All analyses were done using R software (version 3.1.2; http://www.r-project. org/).