血药曲线下面积预测及其模型内部验证方法的比较研究

目的:以口服硫普罗宁为实例,建立血药曲线下面积(AUC)预测模型,比较该模型的内部验证方法。方法:20名健康志愿者单剂量口服硫普罗宁胶囊,以有限抽样法建立多元回归模型预测AUC,用rmse和Pe比较模拟法、Bootstrap法、Jackknife法验证回归模型的效果。结果:模拟数据验证回归模型(AUC0～48=9.36 1.81C3 23.28C8)预测性良好。经模拟法、Bootstrap法、Jackknife法扩大样本量后,估算的回归模型参数分别为Intercept=17.58、9.33、9.84,C3的回归参数M1=1.01、1.76、1.84,C8的回归参数M2=6.17、23.14、21.36,并用原始数据计算其rmse分别为3.87、1.94、2.23。结论:用模拟的c-t数据预测AUC,可用于验证LSS回归模型,但不能通过模拟的c-t数据估算LSS回归模型参数达到这一目的;Bootstrap法和Jackknife法可验证LSS回归模型,结果相近。

Prediction of the AUC and the comparison of approaches for validation

AIM:To predict the AUC with an example by the limited sampling strategy(LSS) and compare the performance of three approaches for internal validation.METHODS:Twenty healthy subjects received 200 mg tiopronin po.A multi-linear regression analysis of the AUC against the tiopronin concentration was carried out to estimate AUC by LSS.According to the values of rmse and Pe,the regression model was validated by the simulation,Bootstrap method and Jackknife method.RESULTS:The model(AUC0-48=9.36 1.81C3 23.28C8) was accurately validated by the simulation data.The parameters obtained from the simulation,Bootstrap method and Jackknife method were that Intercept=17.58,9.33,9.84;M1=1.01,1.76,1.84;M2=6.17,23.14,21.36;and rmse=3.87,1.94,2.23.CONCLUSION:The simulation can evaluate the multi-linear regression model by the estimated AUC,and not by the estimated regression model parameters.The Bootstrap and the Jackknife are appropriate approaches as internal validation of the model.