Multi-Agent Modeling and Analysis of Non-Linear Stochastic Epidemic Dynamics

Date of Award

5-9-2026

Degree Name

M.S. in Mechanical Engineering

Department

Department of Mechanical andAerospace Engineering

Advisor/Chair

Subramanian Ramakrishnan

Abstract

Modeling the spread of infectious diseases requires capturing complex interactions between individuals, spatial heterogeneity, and temporal variability, which are often oversimplified in traditional epidemiological models. Classical approaches, including ordinary and partial differential equation-based frameworks such as the Susceptible–Infectious–Recovered (SIR) model, often rely on assumptions of homogeneous mixing and deterministic behavior, limiting their ability to represent localized transmission dynamics. This study addresses this limitation by developing a data-driven, spatiotemporal agent-based modeling (ABM) framework to simulate epidemic spread at a localized scale. The primary objective of this research is to design, develop, and validate an agent-based model capable of accurately reproducing real-world epidemic dynamics while capturing individual-level interactions and stochastic variability. The model represents individuals as agents with heterogeneous mobility patterns, circadian activity cycles, and exposure-based transmission behavior. Agent movement is modeled using an Active Brownian Motion formulation, and infection probability is determined based on proximity and duration of contact. Model parameters are calibrated using Particle Swarm Optimization (PSO) by minimizing the mean squared error (MSE) between simulated and observed data. The model is validated using real COVID-19 datasets from ZIP code 45207, Cincinnati, Ohio, across three distinct time periods representing varying transmission regimes. Results demonstrate that the proposed framework accurately reproduces both the magnitude and temporal progression of infection and recovery trends, while also capturing short-term fluctuations and localized variability observed in empirical data. Comparative analysis with a PDE-based model shows that while continuum approaches capture overall trends, they fail to effectively represent micro-scale dynamics effectively. In conclusion, the proposed ABM framework provides a robust and realistic approach for modeling epidemic spread by integrating mobility, exposure-based transmission, and data-driven calibration. The findings highlight the importance of discrete modeling approaches for capturing localized epidemic behavior and provide a foundation for future developments in hybrid modeling, intervention analysis, and real-time epidemic forecasting.

Keywords

Mechanical Engineering

Comments

OCLC No. 1591830116

Rights Statement

Copyright 2026, author.

Link to Full Text

Share

COinS
 
 
 

Links