simulate-pk

name: simulate-pk description: PK/PD simulation workflow using rxode2 or mrgsolve. Scaffolds dose scenarios, population simulations, and prediction interval plots.

PK/PD Simulation

Usage

/simulate
/simulate rxode2
/simulate mrgsolve
/simulate <run_number>

What This Does

Scaffolds a PK/PD simulation workflow:

1. Define or load model parameters (from NONMEM run or manual input)
2. Set up dose scenarios
3. Run population simulations with variability
4. Generate spaghetti + median/PI plots
5. Optionally propagate parameter uncertainty

Protocol

Step 1: Fetch Documentation

Use Context7 for the selected engine (rxode2 or mrgsolve).

Step 2: Determine Model Source

If $ARGUMENTS is a run number:

• Load parameters from NONMEM output using xpose/aipharma
• Translate NONMEM model structure to rxode2/mrgsolve syntax

If no run number:

• Ask user for model structure and parameters
• Or use a previously defined model file

Step 3: Define Dose Scenarios

Ask user for:

• Dose levels and regimens
• Number of subjects per scenario
• Simulation duration
• Special populations (if covariate effects included)

Step 4: Scaffold Report

Copy templates/simulation-report.qmd to reports/. Configure for selected engine and scenarios.

Step 5: Generate Simulations

• Set set.seed() for reproducibility
• Run N subjects per scenario (default: 1000)
• Calculate median and 90% prediction intervals
• Generate:
  • Spaghetti plots (individual profiles)
  • Median + PI ribbon plots
  • Dose comparison overlays

Step 6: Review

• Check no negative concentrations
• Verify steady-state achieved if applicable
• Compare to observed data if available

Non-Negotiables

• set.seed() at top of every simulation script
• Named parameter vectors (not positional)
• Document all dose scenarios in header comments
• dpi = 300 for all figures
• Include parameter uncertainty propagation when bootstrap available