BayesianFitForecast: A User-Friendly R Toolbox for Parameter Estimation and Forecasting with Ordinary Differential Equations

TL;DR

BayesianFitForecast is an R toolbox that simplifies Bayesian parameter estimation and forecasting for ODE models, enhancing accessibility for health and epidemiological applications.

Contribution

It provides an easy-to-use interface for Bayesian analysis of ODE models, including automatic Stan file generation and application to real-world epidemiological data.

Findings

Successfully applied to 1918 influenza data in San Francisco.

Demonstrated robustness with simulated time series data.

Offers comprehensive model diagnostics and performance metrics.

Abstract

Background: Mathematical models based on ordinary differential equations (ODEs) are essential tools across various scientific disciplines, including biology, ecology, and healthcare informatics. They are used to simulate complex dynamic systems and inform decision-making. In this paper, we introduce BayesianFitForecast, an R toolbox specifically developed to streamline Bayesian parameter estimation and forecasting in ODE models, making it particularly relevant to health informatics and public health decision-making. The toolbox is available at https://github.com/gchowell/BayesianFitForecast/. Results: This toolbox enables automatic generation of Stan files, allowing users to configure models, define priors, and analyze results with minimal programming expertise. To demonstrate the versatility and robustness of BayesianFitForecast, we apply it to the analysis of the 1918 influenza pandemic in San Francisco, comparing Poisson and negative binomial error structures within the SEIR model. We also test it by fitting multiple time series of state variables using simulated data. BayesianFitForecast provides robust tools for evaluating model performance, including convergence diagnostics, posterior distributions, credible intervals, and performance metrics. Conclusion: By improving the accessibility of advanced Bayesian methods, this toolbox significantly broadens the application of Bayesian inference methods to dynamical systems critical for healthcare and epidemiological forecasting. A tutorial video demonstrating the toolbox's functionality is available at https://youtu.be/jnxMjz3V3n8.