pharmsol

A high-performance Rust library for pharmacokinetic/pharmacodynamic (PK/PD) simulation using analytical solutions, ordinary differential equations (ODEs), or stochastic differential equations (SDEs).

Installation

Add pharmsol to your Cargo.toml, either manually or using

cargo add pharmsol

Quick Start

use pharmsol::*;

// Create a subject with an IV infusion and observations
let subject = Subject::builder("patient_001")
    .infusion(0.0, 500.0, 0, 0.5)  // 500 units over 0.5 hours
    .observation(0.5, 1.645, 0)
    .observation(1.0, 1.216, 0)
    .observation(2.0, 0.462, 0)
    .observation(4.0, 0.063, 0)
    .build();

// Define parameters: ke (elimination rate), v (volume)
let ke = 1.022;
let v = 194.0;

// Use the built-in one-compartment analytical solution
let analytical = equation::Analytical::new(
    one_compartment,
    |_p, _t, _cov| {},
    |_p, _t, _cov| lag! {},
    |_p, _t, _cov| fa! {},
    |_p, _t, _cov, _x| {},
    |x, p, _t, _cov, y| {
        fetch_params!(p, _ke, v);
        y[0] = x[0] / v;  // Concentration = Amount / Volume
    },
    (1, 1),  // (compartments, outputs)
);

// Get predictions
let predictions = analytical.estimate_predictions(&subject, &vec![ke, v]).unwrap();

ODE-Based Models

For custom or complex models, define your own ODEs:

use pharmsol::*;

let ode = equation::ODE::new(
    |x, p, _t, dx, _b, rateiv, _cov| {
        fetch_params!(p, ke, _v);
        // One-compartment model with IV infusion support
        dx[0] = -ke * x[0] + rateiv[0];
    },
    |_p, _t, _cov| lag! {},
    |_p, _t, _cov| fa! {},
    |_p, _t, _cov, _x| {},
    |x, p, _t, _cov, y| {
        fetch_params!(p, _ke, v);
        y[0] = x[0] / v;
    },
    (1, 1),
);

Supported Analytical Models

One-compartment with IV infusion
One-compartment with IV infusion and oral absorption
Two-compartment with IV infusion
Two-compartment with IV infusion and oral absorption
Three-compartment with IV infusion
Three-compartment with IV infusion and oral absorption

Performance

Analytical solutions provide 20-33× speedups compared to equivalent ODE formulations. See benchmarks for details.