Back

Democratizing Infectious Disease Modeling: An AI Assistant for Generating, Simulating, and Analyzing Dynamic Models

Proctor, J. L.; Chabot-Couture, G.

2024-07-17 epidemiology
10.1101/2024.07.17.24310520 medRxiv
Show abstract

Understanding and forecasting infectious disease spread is pivotal for effective public health management. Traditional dynamic disease modeling is an essential tool for characterization and prediction, but often requires extensive expertise and specialized software, which may not be readily available in low-resource environments. To address these challenges, we introduce an AI-powered modeling assistant that utilizes advanced capabilities from OpenAIs latest models and functionality. This tool enhances the accessibility and usability of infectious disease models and simulation frameworks by allowing users to generate or modify model configurations through intuitive natural language inputs or by importing explicit model descriptions. Our prototype integrates with an established open-source disease simulation framework called the Compartmental Modeling Software (CMS) to provide a seamless modeling experience from setup to analysis. The AI assistant efficiently interprets disease model parameters, constructs accurate model files, executes simulations in a controlled environment, and assists in result interpretation using advanced analytics tools. It encapsulates expert knowledge and adheres to best practices to support users ranging from novices to expert modelers. Furthermore, we discuss the limitations of this AI assistant, particularly its performance in complex scenarios where it might generate inaccurate specifications. By enhancing the ease of disease modeling and supporting ongoing capacity-building initiatives, we believe that AI assistants like this one could significantly contribute to global health efforts by empowering researchers, especially in regions with limited resources, to develop and refine their disease models independently. This innovative approach has the potential to democratize disease modeling in global health, offering a scalable solution that adapts to diverse needs across a wide-range of geographies, languages, and populations.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.