What problem does it solve?
Pharmacokinetic and pharmacodynamic analyses are full of silent errors: lambda_z windows chosen by plain r-squared, AUCinf reported when a quarter of it is extrapolated, AIC selecting unidentifiable compartments, and NM-TRAN reading BLQ as a real zero. This Skill provides validated Python scripts and references that compute exposure metrics, fit structural models, and flag exactly these failure modes before they reach a report.
Core Features & Use Cases
- Non-compartmental analysis and compartmental fitting: nca.py derives AUC, Cmax, lambda_z, and half-life with explicit BLQ and lambda_z rules, while fit_compartmental.py compares 1-, 2-, and 3-compartment models with AIC/BIC, F tests, and identifiability diagnostics.
- Regimen simulation, exposure-response, and bioequivalence: simulate_regimen.py projects steady-state attainment across a population, exposure_response.py fits Emax and concentration-QTc models, and bioequivalence.py applies average BE, EMA ABEL, and FDA RSABE criteria with exact power calculations.
- Scaling, DDI, and therapeutic drug monitoring: allometry_and_fih.py handles allometric scaling with maturation and first-in-human dose, ddi_static.py applies ICH M12 basic and mechanistic static models, and tdm_bayes.py performs MAP Bayesian estimation from measured levels.
- Use Case: Given a concentration-time CSV from a single-dose study, run nca.py to get exposure metrics with findings about extrapolated AUC, then fit_compartmental.py to select a defensible structural model with parameter RSEs and residual diagnostics.
Quick Start
Ask the agent to run a non-compartmental analysis on your concentration-time CSV with a stated dose and route, and report the exposure metrics with any findings.
