Julia Documentation¶

Welcome to the Julia documentation for NeoPKPD. The Julia core (NeoPKPD.jl) provides the foundation for all PK/PD modeling capabilities.

Why Julia?¶

NeoPKPD uses Julia as its core language for several reasons:

Performance: Near-native speed through JIT compilation
Mathematical Expressiveness: Natural syntax for differential equations
Ecosystem: World-class ODE solvers via DifferentialEquations.jl
Multiple Dispatch: Flexible, extensible type system

Quick Start¶

using NeoPKPD

# One-compartment IV bolus simulation
params = OneCompIVBolusParams(5.0, 50.0)  # CL=5 L/h, V=50 L
doses = [DoseEvent(0.0, 100.0)]            # 100 mg at t=0
spec = ModelSpec(OneCompIVBolus(), "example", params, doses)

grid = SimGrid(0.0, 24.0, collect(0.0:1.0:24.0))
solver = SolverSpec(:Tsit5, 1e-10, 1e-12, 10_000_000)

result = simulate(spec, grid, solver)
println(result.observations[:conc])

Documentation Sections¶

Tutorial

Step-by-step introduction to NeoPKPD Julia API

Start Tutorial
Models

Complete reference for PK and PD models

Models Reference
Population Modeling

IIV, IOV, covariates, and residual error

Population
NCA

Non-compartmental analysis

NCA Reference
Parameter Estimation

FOCE-I, SAEM, and Laplacian methods

Estimation
Visual Predictive Check

VPC, pcVPC, and stratification

VPC Reference
Model Import

NONMEM and Monolix file parsing

Import Reference
Clinical Trials

Trial simulation and power analysis

Trial Reference

Core Types¶

ModelSpec¶

The central type for defining a simulation:

struct ModelSpec{M<:ModelKind, P<:AbstractParams}
    model::M           # Model type (e.g., OneCompIVBolus)
    name::String       # Simulation identifier
    params::P          # Model parameters
    doses::Vector{DoseEvent}
end

SimGrid¶

Defines the time domain:

struct SimGrid
    t0::Float64        # Start time
    t1::Float64        # End time
    saveat::Vector{Float64}  # Output time points
end

SolverSpec¶

Configures the ODE solver:

struct SolverSpec
    alg::Symbol        # Algorithm (:Tsit5, :Rosenbrock23, etc.)
    reltol::Float64    # Relative tolerance
    abstol::Float64    # Absolute tolerance
    maxiters::Int      # Maximum iterations
end

SimResult¶

Simulation output:

struct SimResult
    t::Vector{Float64}
    states::Dict{Symbol, Vector{Float64}}
    observations::Dict{Symbol, Vector{Float64}}
    metadata::Dict{Symbol, Any}
end

Available Models¶

Pharmacokinetic Models¶

Model Type	Parameters	Description
`OneCompIVBolus`	CL, V	IV bolus, first-order elimination
`OneCompOralFirstOrder`	Ka, CL, V	Oral with first-order absorption
`TwoCompIVBolus`	CL, V1, Q, V2	Two-compartment IV
`TwoCompOral`	Ka, CL, V1, Q, V2	Two-compartment oral
`ThreeCompIVBolus`	CL, V1, Q2, V2, Q3, V3	Three-compartment IV
`TransitAbsorption`	N, Ktr, Ka, CL, V	Transit compartment absorption
`MichaelisMentenElimination`	Vmax, Km, V	Saturable elimination

Pharmacodynamic Models¶

Model Type	Parameters	Description
`DirectEmax`	E0, Emax, EC50	Direct effect model
`SigmoidEmax`	E0, Emax, EC50, gamma	Hill equation
`BiophaseEquilibration`	ke0, E0, Emax, EC50	Effect compartment
`IndirectResponseTurnover`	Kin, Kout, R0, Imax, IC50	Indirect response

Key Functions¶

Simulation¶

# Single subject simulation
result = simulate(spec::ModelSpec, grid::SimGrid, solver::SolverSpec)

# Population simulation
result = simulate_population(pop_spec::PopulationSpec, grid, solver)

Parameter Estimation¶

# FOCE-I estimation
result = estimate(data, spec, FOCEConfig())

# SAEM estimation
result = estimate(data, spec, SAEMConfig())

NCA¶

# Non-compartmental analysis
result = run_nca(times, conc, dose; config=NCAConfig())

VPC¶

# Visual predictive check
vpc_result = compute_vpc(observed, simulated; config=VPCConfig())