Skip to content

Model Import

NeoPKPD can import models from other pharmacometrics platforms, enabling migration and interoperability.

Supported Formats

NONMEM Import

Supported Features

Feature Support Notes
ADVAN1 (1-comp IV) ✅ Full Direct mapping
ADVAN2 (1-comp oral) ✅ Full Ka, CL, V
ADVAN3 (2-comp IV) ✅ Full CL, V1, Q, V2
ADVAN4 (2-comp oral) ✅ Full Ka, CL, V1, Q, V2
ADVAN11 (3-comp IV) ✅ Full Full three-comp
ADVAN12 (3-comp oral) ✅ Full With absorption
TRANS parameters ✅ Full TRANS1-4
$THETA ✅ Full Fixed effects
$OMEGA ✅ Full Random effects
$SIGMA ✅ Full Residual error
$ERROR ⚠️ Partial Common patterns
$PK ⚠️ Partial Standard code

Quick Start

using NeoPKPD

# Import NONMEM control stream
model = import_nonmem("run001.ctl")

# Access parsed components
println("Model type: ", model.advan)
println("THETA: ", model.theta)
println("OMEGA: ", model.omega)

# Convert to NeoPKPD ModelSpec
spec = to_model_spec(model)

CLI Usage

./bin/neopkpd import --input run001.ctl --format nonmem --out model.json

Monolix Import

Supported Features

Feature Support
Structural models ✅ Full
Parameter definitions ✅ Full
Random effects ✅ Full
Covariate models ⚠️ Partial
Error models ✅ Full

Quick Start

# Import Monolix project
model = import_monolix("project.mlxtran")

# Convert to NeoPKPD
spec = to_model_spec(model)

CDISC Data Import

Supported Domains

Domain Description Support
PC Pharmacokinetic Concentrations ✅ Full
EX Exposure (Dosing) ✅ Full
DM Demographics ✅ Full
VS Vital Signs ⚠️ Partial
LB Laboratory ⚠️ Partial

Quick Start

# Import CDISC data
data = import_cdisc(
    pc_path = "pc.csv",
    ex_path = "ex.csv",
    dm_path = "dm.csv",
    format = :csv           # or :xpt for SAS transport
)

# Access standardized data
println("Subjects: ", length(unique(data.usubjid)))
println("Observations: ", length(data.dv))

Data Structure

struct CDISCData
    # Subject identifiers
    usubjid::Vector{String}
    subjid::Vector{Int}

    # Time and observations
    times::Vector{Float64}
    dv::Vector{Float64}

    # Dosing
    doses::Vector{DoseEvent}
    dose_times::Vector{Float64}
    dose_amounts::Vector{Float64}

    # Demographics
    age::Vector{Float64}
    sex::Vector{Symbol}
    weight::Vector{Float64}

    # Metadata
    study::String
    analyte::String
end

Migration Workflow

NONMEM to NeoPKPD

graph LR
    A[NONMEM .ctl] --> B[import_nonmem]
    B --> C[ParsedModel]
    C --> D[to_model_spec]
    D --> E[NeoPKPD ModelSpec]
    E --> F[Simulate/Estimate]

Step-by-Step

  1. Parse control stream 

    parsed = import_nonmem("run001.ctl")

  2. Review parsed model 

    println("ADVAN: ", parsed.advan)
println("Parameters: ", parsed.theta)
println("Random effects: ", parsed.omega)

  3. Convert to NeoPKPD 

    spec = to_model_spec(parsed)

  4. Validate with simulation 

    result = simulate(spec, grid, solver)

  5. Compare to NONMEM output 

    # Load NONMEM predictions
nm_pred = load_nonmem_output("run001.tab")

# Compare
@assert isapprox(result.observations[:conc], nm_pred.IPRED, rtol=0.01)

Limitations

NONMEM

  • Custom \(PK/\)ERROR blocks may not parse correctly
  • User-defined ADVAN/TRANS not supported
  • Complex conditional logic requires manual conversion

Monolix

  • Custom structural models need adaptation
  • Some advanced features may not translate

Workarounds

For unsupported features:

  1. Simplify the NONMEM code before import
  2. Manually define the NeoPKPD model
  3. Use the CLI --validate flag to check import accuracy

Next Steps