MATHEMATICAL-MODELING OF HUMAN GROWTH - A COMPARATIVE-STUDY
MATHEMATICAL-MODELING OF HUMAN GROWTH - A COMPARATIVE-STUDY
复制标题
DOI：
10.1002/ajhb.1310040112
复制
发表时间：
1992-01-01
影响因子：
2.9
通讯作者：
CHUMLEA, WC
中科院分区：
文献类型：
--
作者：
GUO, SM;SIERVOGEL, RM;CHUMLEA, WC
核回归是一种非参数过程，可以很好地逼近单个序列数据。当参数方法由于对曲线形状的限制假设而不适合时，该方法是有用的和灵活的。在本研究中，我们比较了核回归与两种模型拟合人类身高增长的关系，其中一种模型包含了可能存在的生长中期突增，而另一种模型没有。两组数学函数和非参数核回归拟合了227名参加费尔斯纵向研究的参与者的身高序列测量。描述生长突增的时间、幅度和持续时间的生长参数，如生长中期突增和青春期突增，是从每个参与者的拟合模型和核回归中得出的。比较了两种参数模型和核回归的总体拟合优度以及量化生长事件的时间、增长率和持续时间的能力。Preece-Baines模型没有描述中期增长突增。从Preece-Baines模型导出的生长参数显示，青春期爆发的开始时间较早，持续时间较长，速度增长较慢。与三重逻辑模型相比，带宽为2年的核回归和二阶多项式核函数的拟合效果相对较好。在核回归和三重逻辑模型之间推导出的青春期爆发的生物学参数是相似的。核回归估计中期爆发的开始时间更早，速度的增加速度更快。
Kernel regression is a nonparametric procedure that provides good approximations to individual serial data. The method is useful and flexible when a parametric method is inappropriate due to restricted assumptions on the shape of the curve. In the present study, we compared kernel regression in fitting human stature growth with two models, one of which incorporates the possible existence of the midgrowth spurt while the other does not. Two families of mathematical functions and a nonparametric kernel regression were fitted to serial measures of stature on 227 participants enrolled in the Fels Longitudinal Study. The growth parameters that describe the timing, magnitude, and duration of the growth spurt, such as midgrowth spurt and pubertal spurt, were derived from the fitted models and kernel regression for each participant. The two parametric models and kernel regression were compared in regard to their overall goodness of fit and their capabilities to quantify the timing, rate of increase, and duration of the growth events. The Preece-Baines model does not describe the midgrowth spurt. The derived growth parameters from the Preece-Baines model show an earlier onset and a longer duration of the pubertal spurt, and a slower increase in velocity. The kernel regression with bandwidth 2 years and a second-order polynomial kernel function yields relatively good fits compared with the triple logistic model. The derived biological parameters for the pubertal spurt are similar between the kernel regression and the triple logistic model. Kernel regression estimates an earlier onset and a more rapid increase of velocity for the midgrowth spurt.