Cross-Entropy Approximate Bayesian Computation

CE-ABC: Cross-Entropy Approximate Bayesian Computation is a Matlab package that implements a framework for uncertainty quantification in mechanistic epidemic models defined by ordinary differential equations (ODEs). This package combines the cross-entropy method for optimization and approximate Bayesian computation for statistical inference. With straightforward adaptations, the CE-ABC strategy can be applied to various other systems, including mechanical, electrical, and coupled systems.

Table of Contents

• Overview
• Features
• Usage
• Documentation
• Reproducibility
• Authors
• Citing CE-ABC
• License
• Institutional support
• Funding

Overview

CE-ABC addresses model calibration and uncertainty quantification in mechanistic models, primarily for epidemic modeling. The package integrates the cross-entropy method, which is a powerful optimization technique, with approximate Bayesian computation, a statistical inference method. This combination allows for efficient and accurate calibration and uncertainty quantification in ODE-based models.

For more details, refer to the following paper:

• A. Cunha Jr, D. A. W. Barton, and T. G. Ritto, Uncertainty quantification in mechanistic epidemic models via cross-entropy approximate Bayesian computation, Nonlinear Dynamics, vol. 111, pp. 9649–9679, 2023. DOI

Preprint available here.

Features

• Combines cross-entropy method for optimization with approximate Bayesian computation for statistical inference
• Applicable to mechanistic models defined by ODEs
• Flexible framework for various systems (mechanical, electrical, coupled, etc.)
• Numerically robust and efficient implementation
• Educational style for intuitive use
• Includes example scripts for representative benchmark tests

Usage

To get started with CE-ABC, follow these steps:

1. Clone the repository:

   git clone https://github.com/americocunhajr/CE-ABC.git
   

2. Navigate to the code directory:

   cd CE-ABC/CE-ABC-1.0
   

3. For a deterministic simulation with SEIRpAHD model, execute:

   Main_IVP_SEIRpAHD
   

4. For a stochastic simulation with SEIRpAHD model, execute:

   Main_CE_ABC_SEIRpAHD
   

5. For a stochastic simulation with SEIRpAHDbeta model, execute:

   Main_CE_ABC_SEIRpAHDbeta
   

6. To plot Rio de Janeiro COVID-19 data, execute:

   Main_COVID19RJ_Data_plot
   

Documentation

CE-ABC routines are well-commented to explain their functionality. Each routine includes a description of its purpose and a list of inputs and outputs. Examples with representative benchmark tests are provided to illustrate the code’s functionality.

Reproducibility

Simulations done with CE-ABC are fully reproducible, as can be seen on this CodeOcean capsule.

Authors

• Americo Cunha Jr
• David A. W. Barton
• Thiago G. Ritto

Citing CE-ABC

If you use CE-ABC in your research, please cite the following publication:

• A. Cunha Jr, D. A. W. Barton, and T. G. Ritto, Uncertainty quantification in mechanistic epidemic models via cross-entropy approximate Bayesian computation, Nonlinear Dynamics, vol. 111, pp. 9649–9679, 2023 https://doi.org/10.1007/s11071-023-08327-8

@article{CunhaJr2023p9649,
   author  = {A {Cunha~Jr} and D. A. W. Barton and T. G. Ritto},
   title   = {Uncertainty quantification in mechanistic epidemic models via cross-entropy approximate Bayesian computation},
   journal = {Nonlinear Dynamics},
   year    = {2023},
   volume  = {111},
   pages   = {9649–9679},
   doi    = {10.1007/s11071-023-08327-8},
}

License

CE-ABC is released under the MIT license. See the LICENSE file for details. All new contributions must be made under the MIT license.

Institutional support

       

Funding