我在做必需氨基酸需要量方面的一个试验，现在数据统计分析上碰到一个难题，就是相关国外文献都是用一个“two-phase linear regression crossover model”的统计模型来确定两条直线的拐点（breakpoint)，从而得出该氨基酸的最低生理需要量。我也看了几本有关回归分析的SAS书籍，可是没有相关内容，更没有现成的程序模板，希望能得到高手的帮助！
相关背景：

This model selects for the minimum residual SE in a stepwise partitioning of data points between 2 regression lines. The first regression line has a slope and the second line is horizontal with minimal or no slope. The mixed procedure takes into account the correlations among observations made on the same participant and the possible heterogeneous variances among the measurements made on the same participant over time and is well suited for repeated-measures study designs (25). Different variancecovariance structures (simple, compound symmetry, unstructured, first order auto-regressive, and first order ante-dependence) were compared using the information criteria (Akaike Information Criteria, the finite sample corrected Akaike Information Criteria, and the Schwarz’s Bayesian Information Criteria). The final model that best fit the data with the lowest SE, lowest root mean square error, and highest r2 identified the breakpoint estimate or requirement for lysine. The safe
intake of lysine (upper 95% CI, equivalent to the Recommended Dietary Allowance) was calculated using Fieller’s theorem (26). Briefly, 95% CI = breakpoint 6 tdf,a/2 3 SE, where SE is the SE of the breakpoint, df is associated with the residual mean square of the best fit model, and a is the 95% confidence level. Upper 95% CI was calculated using SAS (SAS/STAT version 8.2, SAS Institute).