Raw data and results for the paper "Full conditional non-parametric bootstrap - an evaluation with unbalanced designs and high residual variability"

This depot contains raw data and results for the paper "Full conditional non-parametric bootstrap - an evaluation with unbalanced designs and high residual variability" Data Original Emax and Hill model scenarios can be found in the archive comets_condBoot_data.zip located in the Zenodo depot: https://zenodo.org/records/4059718 Direct link to downloading the datasets: https://zenodo.org/records/4059718/files/comets_condBoot_data.zip?download=1 The data simulated for the other two scenarios are available in the present depot in the Zip file: newDesigns_simulatedData.zip - Unbalanced designs - comb_dropout: mix of patients receiving a full set of doses and others receving low doses - comb_end: mix of patients with rich and sparse designs, with all patients receiving the first and last dose - comb_latestart: mix of patients receiving a full set of doses and others receving high doses - comb_low: mix of patients receiving four or two doses - Increased error - sigma30: 30% residual error - sigma50: 50% residual error Results - Original designs - results for full cNP - results for the other bootstraps can be downloaded from the previous depot: https://zenodo.org/records/4059718 - Unbalanced designs - comb_dropout: mix of patients receiving a full set of doses and others receving low doses - comb_end: mix of patients with rich and sparse designs, with all patients receiving the first and last dose - comb_latestart: mix of patients receiving a full set of doses and others receving high doses - comb_low: mix of patients receiving four or two doses - Increased error - sigma30: 30% residual error - sigma50: 50% residual error Results include empirical SE, bias, RMSE (relative and absolute), parameter estimates, and coverage rates. Running bootstrap with saemix Finally, the Rstudio notebook shows how to run saemix and the different bootstraps on one of the demo datasets available in saemix. Packages (including saemix) needed to run this code are listed at the beginning of the notebook and the PDF shows the results of running the notebook.本数据存储库收录论文《全条件非参数自助法（Full conditional non-parametric bootstrap）——一项针对非均衡设计与高残差变异的评估》的原始数据与研究结果。 原始Emax模型与Hill模型场景的相关数据，可于Zenodo数据存储库的归档文件comets_condBoot_data.zip中获取，链接为：https://zenodo.org/records/4059718；数据集直接下载链接为：https://zenodo.org/records/4059718/files/comets_condBoot_data.zip?download=1。 其余两类仿真场景的数据，存于本存储库的压缩包newDesigns_simulatedData.zip中，具体场景分类如下： - 非均衡设计： - comb_dropout：接受完整剂量疗程与低剂量疗程的患者混合队列 - comb_end：包含密集采样与稀疏采样设计的患者队列，所有受试者均接受首剂与末剂给药 - comb_latestart：接受完整剂量疗程与高剂量疗程的患者混合队列 - comb_low：接受4剂或2剂给药的患者混合队列 - 误差增强设置： - sigma30：30%残差误差 - sigma50：50%残差误差 研究结果部分： - 原始设计相关结果：全条件非参数自助法与其他自助法的运行结果，可从前述Zenodo存储库（https://zenodo.org/records/4059718）下载 - 非均衡设计与误差增强的结果分类与前述数据场景一致，此处不再重复罗列 本次研究的结果包含经验标准误（empirical SE）、偏差（bias）、均方根误差（RMSE，含相对与绝对两种形式）、参数估计值与覆盖率。 关于使用saemix运行自助法：本存储库附带的RStudio笔记本演示了如何针对saemix内置的一则演示数据集，运行saemix与各类自助法。运行该代码所需的依赖包（含saemix）已列于笔记本开篇，配套的PDF文件则展示了该笔记本的完整运行结果。