论文部分内容阅读
目的 寻找适合于拟合各种人体测量指标的生长标准曲线的方法和途径 .方法 采用损失函数 (loss function)和残差平方和 (residual sum of square)对加权三次样条法(weighted cubic smoothing spline)和 λ-中位数 -变异系数法 (λ-median- coefficient of variation,L MS)进行对比研究 .结果 计算了两种方法拟合各组的身高 (height)、体质量 (bodymass)、体块指数 (body m ass index,BMI)的损失函数值和残差平方和 ,给出了两种方法拟合的百分位数曲线 .结论 L MS法简单易实现 ,拟合的百分位数曲线光滑 ,拟合误差小 .L MS法比加权三次样条更适合于拟合身高、体质量、体块指数百分位数曲线
Objective To find a suitable method and approach for fitting the growth standard curve of various anthropometric indicators. The method uses loss function and residual sum of square for weighted cubic smoothing spline. Comparing the λ-median-coefficient of variation (LMS) method. The results calculated two methods to fit the height, body mass, and volume of each group. The loss function value of the body index index (BMI) and the sum of the residual squares give a percentile curve fitted by the two methods. Conclusion The L MS method is simple and easy to implement, and the percentile of the fitting is The curve is smooth and the fitting error is small. The LMS method is more suitable than the weighted cubic spline for fitting the height, body mass, body mass index percentile curve.