BioEq

A Python package for performing bioequivalence calculations and analysis.

Overview

BioEq is a comprehensive Python package for analyzing pharmacokinetic data in bioequivalence studies. It provides methods for processing and analyzing data from both crossover and parallel design bioequivalence studies.

Features

• Crossover 2x2 Design Analysis:

  • Calculation of AUC, Cmax, Tmax, and other PK parameters
  • ANOVA and mixed effects model analysis
  • Point estimates and 90% confidence intervals for bioequivalence assessment

• Parallel Design Analysis:

  • Calculation of PK parameters for parallel group studies
  • ANOVA and t-test analysis
  • Point estimates and 90% confidence intervals

• Replicate Crossover Design Analysis:

  • Support for both partial (3-way) and full (4-way) replicate designs
  • Within-subject variability assessment for reference formulation
  • Reference-scaled average bioequivalence (RSABE) for highly variable drugs
  • Expanded bioequivalence limits based on reference variability

• Additional Features:

  • Estimation of elimination half-life
  • Extrapolation of AUC to infinity
  • Summary statistics for PK parameters
  • Data visualization tools
  • Flexible data input/output handling

Installation

pip install bioeq

Quick Start

import polars as pl
from bioeq import Crossover2x2

# Load your data
data = pl.read_csv("your_data.csv")

# Initialize analyzer
analyzer = Crossover2x2(
    data=data,
    subject_col="SubjectID",
    seq_col="Sequence",
    period_col="Period",
    time_col="Time",
    conc_col="Concentration",
    form_col="Formulation"
)

# Calculate point estimate and confidence intervals
results = analyzer.calculate_point_estimate("log_AUC")
print(f"Point Estimate: {results['point_estimate']:.2f}%")
print(f"90% CI: {results['lower_90ci']:.2f}% - {results['upper_90ci']:.2f}%")

Replicate Crossover Example

import polars as pl
from bioeq import ReplicateCrossover

# Load data for a partial replicate (3-way) design
data = pl.read_csv("replicate_data.csv")

# Initialize analyzer for a partial replicate design
analyzer = ReplicateCrossover(
    data=data,
    design_type="partial",  # Use "full" for 4-way replicate design
    subject_col="SubjectID",
    seq_col="Sequence",
    period_col="Period",
    time_col="Time",
    conc_col="Concentration",
    form_col="Formulation"
)

# Calculate within-subject CV for the reference product
cv_results = analyzer.calculate_within_subject_cv("log_AUC")
print(f"Within-subject CV for reference: {cv_results['cv_percent']:.2f}%")

# If drug is highly variable (CV ≥ 30%), run reference-scaled approach
if cv_results['cv_percent'] >= 30:
    rsabe_results = analyzer.run_rsabe("log_AUC")
    print(f"RSABE criterion met: {rsabe_results['rsabe_criterion_met']}")
    print(f"Expanded BE limits: {rsabe_results['lower_scaled_limit']:.2f}% - {rsabe_results['upper_scaled_limit']:.2f}%")
else:
    # Use standard bioequivalence approach
    print("Standard bioequivalence assessment recommended (CV < 30%)")

Documentation

Comprehensive documentation is available in the docs directory:

• Regulatory Documentation:

  • Validation Strategy
  • Functional Requirements
  • Traceability Matrix
  • Algorithmic Specifications
  • Regulatory References

• Developer Documentation:

  • Development Workflows
  • Notebook Integration

• Bioequivalence Notes:

  • Detailed Notes
  • Concise Summary

For a complete overview of the documentation, see the docs README.

Validation and Regulatory Compliance

BioEq is designed with regulatory requirements in mind. The package includes:

• Validation Module: Built-in functionality to validate calculations against known values
• Traceability: Documentation mapping requirements to implementation
• Algorithm Documentation: Detailed descriptions of implemented methods with scientific references
• Regulatory References: Documentation linking to relevant FDA guidance documents

Running Validation Tests

To validate the package calculations:

# Run the built-in validation suite
bioeq validate --output validation_report.json

Requirements

• Python ≥ 3.10
• Polars ≥ 1.18.0
• NumPy ≥ 2.2.1
• SciPy ≥ 1.15.0
• Statsmodels ≥ 0.14.4
• PyArrow ≥ 19.0.0

License

This project is licensed under the MIT License - see the LICENSE file for details.