Refua Clinical

Simulate virtual patients and clinical trials using PK/PD models. Compare adaptive trial designs and prepare draft protocols for your team to review.

View on GitHub Back to Ecosystem
PK/PD Simulation
Bayesian Adaptive Design
4 Estimand Types
11 Research Citations

What Is Refua Clinical?

Refua Clinical is a simulation package for PK/PD-driven clinical trial design. It generates copula-based virtual patients, runs adaptive multi-arm trial simulations with Bayesian response-adaptive allocation, and produces a tailored protocol that can be re-simulated with modified inputs.

The package directly integrates with the core Refua toolkit, accepting ADMET profiles and affinity predictions as inputs to inform simulation parameters automatically.

Key Capabilities

  • Population PK/PD simulation with inter-individual variability
  • Virtual patient generation using Gaussian copulas
  • Estimand-aware analysis (treatment policy, hypothetical, composite, while-on-treatment)
  • Bayesian response-adaptive allocation with early stopping
  • Dynamic external-control borrowing with discounting
  • Multi-objective design optimization with Pareto front outputs

Tools for Comparing Trial Designs

👥

Virtual Patients

Generate realistic virtual patient populations using Gaussian copulas with configurable marginals and covariate effects for each trial arm.

📐

Protocol Recommendation

Score candidate designs on simulated operating characteristics and expected cost to recommend optimal sample sizes and interim cadences.

🎯

Multi-Objective Optimization

Compare designs that make different trade-offs between statistical power, cost, and duration.

💰

Value-of-Information

Run VOI scenarios to determine whether additional enrollment is worth the investment before committing to expansion decisions.

🌍

Transportability

Assess covariate shift, overlap scores, and risk levels for extrapolating results from reference to target populations.

💡

Explainable Advice

Summarize simulation results, interim decisions, and suggested next steps for your team to review.

End-to-End Workflow

Configure

Initialize a trial config with init-config or accept ADMET inputs from Refua.

Simulate

Run the PK/PD simulation with virtual patients, adaptive allocation, and early stopping rules.

Protocol

Generate a tailored protocol recommendation with optimal sample size and interim schedule.

Optimize

Run multi-objective optimization and VOI analysis to refine the design.

Advise

Summarize the results and explain the recommended next steps.

Biologics & Small Molecules

Refua Clinical supports both small molecule and biologics modalities. Biologics mode adds IV/SC route support, interval dosing, and optional TMDD-like clearance scaling for antibody and protein therapeutics.

The ADMET-aware simulation path automatically adjusts PK parameters based on Refua ADMET profiles, connecting molecular predictions with trial planning assumptions.

Supported Modalities

  • Small molecules with oral PK models
  • Biologics with IV/SC administration routes
  • Interval dosing (q14d, q28d, etc.)
  • TMDD-like clearance scaling
  • ADMET-informed PK parameter adjustment

Get Started

Installation

pip install refua-clinical

With ADMET integration:

pip install "refua-clinical[admet]"

Run a Trial Workup

refua-clinical workup \
  --config config.yaml \
  --output-dir artifacts/full_workup \
  --include-sensitivity

Python API

from refua_clinical import ClinicalStudy

study = (
    ClinicalStudy.default()
    .trial(
        trial_id="oncology-demo",
        indication="Oncology",
        phase="Phase II",
        replicates=96,
    )
    .set("enrollment.total_n", 220)
)

run = study.simulate()
protocol = run.recommend_protocol(replicates_per_candidate=30)
print(run.summary["power"], protocol.protocol["protocol_id"])

Design Better Trials

Compare trial designs, test assumptions, and review simulated outcomes before enrolling patients.

View on GitHub Back to Ecosystem