Overview | Setup | Pulling data from CvTdb | Minimal PK object | Running all possible fitting options | Characterizing the data | Number of data groups with both blood and plasma data available | Number of data groups where time scaling could occur | How many experiments (Chemical, Species, Reference, Route, Media) were flagged for potential dose dependence? | Evaluating time and concentration ranges | Evaluating data variability | Figure 3 | Supplemental Figure 3: Histogram of data variability by time | Evaluating fitting options | Winning model tally | RMSLE for Cmax and AUC (with some tallys for extreme values) | AUC_inf and Cmax comparison | Goodness of fit metrics | Get all predictions | Factor of two model error | Evaluate the various GOF metrics | Rank fitting options | Supplemental Table 2: Save evaluation results | Plots for evaluation of fitting options | Goodness-of-fit metrics Rsq and RMSLE across fitting options | Cmax RMSLE vs. AUC RMSLE across fitting options | Analysis Plots and Tables for Best Set of Fitting Options | Supp Figure 7: invivoPKfit model performance | Figure 4 and Supplemental Figures 4 & 5: Model performance vs Data Variability | Figure 5: Multiple goodness-of-fit metrics validate model performance | Figure 7: Examples fits for chemicals with R-squared and within 2-fold | Supp. Fig 8: plots of the five data groups that were best fit by the null model | Figure 6: Comparing derived TK stats with human TK stats from Lombardo et al. | Supp. Fig. 9: Fgutabs comparison with literature values compiled by Musther et al. (2014) | Supp. Fig. 10: Parallelization decreases runtime of invivoPKfit | Summary fit data for all fitting options, all models, all data groups | Individual study level analysis | All plots for best set of fitting options | Print sessionInfo()

Introduction | The model fitting workflow | Select data to be fit | Initialize a pk object | Step 1: Pre-processing data | Step 2: Data information | Step 3: Pre-fitting | Step 4: Model fitting | Fast-forward to model fitting | Post-fitting: Getting information about the fitted models | Coefficients, residuals, and predictions | Plots | Model evaluation metrics | Data | Toxicokinetic statistics | Winning model | Anatomy of a pk object | Original data | Variable mapping | Expressions in the aes() mapping specification | Current status | Data grouping | Settings for data pre-processing | routes_keep | media_keep | ratio_conc_dose | impute_loq, loq_group and calc_loq_factor | impute_sd, sd_group | Data information settings | Optimizer settings | Data transformations (scalings) | Concentration transformations (scalings) | dose_norm | log10_trans | my_pk$scales$time | Models to be fitted | Error model to apply during fitting | "Hierarchical" error model | "Pooled" analysis | "Separate" analysis | Providing new instructions for a pk object | Checking the current instructions for a pk object | Check original data | Check mapping | Check status | Check data grouping | Check settings_preprocess | Check nca_group | Check settings_optimx | Check scale_conc | Check scale_time | Check stat_model | Check stat_error_model | Replacing instructions with new ones | What happens to a pk object as you go through the steps of model fitting workflow | Initialization | Pre-processing data | Data summary info | Pre-fitting | Fitting | Keeping track of workflow status as you change instructions | A few worked examples | Compare fits made using different error models | How to add a new PK model to invivopkfit | Model function requirements | conc_fun requirements | auc_fun requirements | params_fun requirements | tkstats_fun requirements | Simplified model modification interface | Caveats

Initialized the Random Number Generator: | Do the noncompartment model fit | Do the 2-compartment model fit | Plot the 2-compartment concentration vs. time | Do the 1-compartment model fit | Plot one compartment concentration vs. time | Output the dataset analyzed: