Documentation

Learn how to use ModelBuilder for population pharmacokinetic and pharmacodynamic modeling.

Quick Start Guide

ModelBuilder makes population PK/PD modeling accessible. Follow these steps to fit your first model:

  1. 1
    Prepare your data — Format your dataset as CSV with required columns: ID, TIME, DV (observations), AMT (doses), and EVID (event type).
  2. 2
    Create a project — Click "New Project", name it, and upload your CSV file.
  3. 3
    Map columns — Assign roles to your dataset columns (ID, TIME, DV, etc.).
  4. 4
    Configure model — Select compartments, administration route, and estimation method.
  5. 5
    Run & review — nlmixr2 fits your model. Review parameters, diagnostics, and download your report.

Data Format Requirements

Required Columns

ColumnDescriptionExample
IDSubject identifier1, 2, 3...
TIMETime since first dose (hours)0, 1, 2, 4, 8...
DVDependent variable (concentration)15.2, 8.4, 3.1...
AMTDose amount (mg)100, 0, 0...
EVIDEvent ID (0=obs, 1=dose)1, 0, 0...

💡 Tip

If your data uses different column names, ModelBuilder will auto-detect common alternatives (e.g., SUBJ for ID, CONC for DV). You can also manually map any column.

Understanding Model Types

1-Compartment Model

Simplest model assuming drug distributes instantly throughout the body. Best for drugs with rapid distribution.

Parameters: CL (clearance), V (volume), Ka (if oral)

2-Compartment Model

Accounts for distribution between central and peripheral compartments. Most commonly used for IV drugs and many oral drugs.

Parameters: CL, V1 (central), V2 (peripheral), Q (intercompartmental CL)

3-Compartment Model

Adds a deep peripheral compartment for drugs with complex distribution (e.g., highly lipophilic drugs, some biologics).

Parameters: CL, V1, V2, V3, Q2, Q3

SAEM Algorithm

Stochastic Approximation Expectation-Maximization (SAEM) is our recommended estimation method for most models. It's robust, handles complex models well, and provides reliable parameter estimates even with sparse data.

When to use SAEM:

  • Default choice for most population PK models
  • Models with multiple random effects
  • Sparse or unbalanced data designs
  • Non-linear mixed effects models

Ready to start modeling?

Create your first project and fit a population PK model in minutes.

Create New Project