Model Evaluation

Usage

/model-eval <run_number>
/model-eval 101

What This Does

1. •Check run status -- convergence, covariance, boundary warnings
2. •Extract parameters -- estimates, RSE, IIV (CV%), shrinkage
3. •Generate GOF plots -- DV vs PRED, DV vs IPRED, CWRES vs TIME, CWRES vs PRED
4. •Report OFV and condition number
5. •Flag issues -- THETA=0, high RSE, high shrinkage, failed steps

Protocol

Step 1: Locate Run

run_dir <- file.path("models", paste0("run", run_number))
lst_file <- file.path(run_dir, paste0("run", run_number, ".lst"))

Step 2: Check Status

library(aipharma)
status <- check_run_status(run_number, run_dir = "models")

If convergence OR covariance failed, report immediately and flag.

Step 3: Load xpose Data

library(xpose)
xpdb <- xpose_data(runno = run_number, dir = run_dir)

Step 4: Parameter Summary

params <- get_parameter_estimates_xpose(xpdb)

Present as table: Parameter | Estimate | RSE(%) | IIV CV(%) | Shrinkage(%)

Step 5: GOF Plots

Generate using xpose with type = "ps" (points + smooth):

# Standard 4-panel GOF
dv_vs_pred(xpdb, type = "ps")
dv_vs_ipred(xpdb, type = "ps")
res_vs_idv(xpdb, res = "CWRES", type = "ps")
res_vs_pred(xpdb, res = "CWRES", type = "ps")

Step 6: Report

Present structured evaluation:

• •Run number and description
• •Convergence: PASS/FAIL
• •Covariance: PASS/FAIL
• •OFV value
• •Parameter table
• •Issues flagged
• •GOF plot interpretations
• •Recommendation: Accept / Revise / Reject

Non-Negotiables

• •NEVER hardcode parameter values
• •ALWAYS check convergence before extracting parameters
• •GOF plots MUST use type = "ps"
• •Log every evaluation in model development log