Mathematical Models of Disease Dynamics: A Deterministic and Stochastic Framework
Date Issued
May 2026
Author(s)
Young, Amber A
The University of Tennessee
Advisor(s)
Suzanne Lenhart
Additional Advisor(s)
Rebecca T. Trout Fryxell
Olivia Feldman
Christopher Strickland
Abstract
In this thesis, we present two projects. La Crosse Virus (LACV) is a mosquito-borne disease that is spread via the bite of an infected mosquito, with most reported cases occurring in individuals under 20 years old. Individuals with LACV may also develop La Crosse Encephalitis, but most cases are asymptomatic and go unreported, and in recent years, counties in North Carolina and Tennessee have been identified as hotspots for the virus. Humans serve as dead-end hosts and therefore do not play a role in LACV
transmission. Using a system of ordinary differential equations, we developed the first model that accounts for inter- and intra-year variability of LACV transmission in mosquito and human populations in western North Carolina and eastern Tennessee. Environmental effects on mosquito populations were incorporated through non-decreasing, time-varying functional forms of precipitation and temperature. Human demographics were parameterized using population data, while yearly case data and past seroprevalence studies were used to estimate human transmission rates. Klebsiella pneumoniae (K. pneumoniae) is a Gram-negative bacterium in the family Enterobacteriaceae, and one cause of healthcare-acquired infections. Although K. pneumoniae can be found in contaminated water, soil, and medical equipment, but transmission occurs most commonly through person-to-person contact. We developed a continuous-time, stochastic model at the ward-level that incorporates patient-level hospital movement data between wards to investigate the intrahospital transmission dynamics of K. pneumoniae. We have de-identified patient data from a hospital located in St. Louis, Missouri, including unique patient identifiers, admission and discharge times, and ward locations. Diagnostic culture data for a subset of patients, including the time, culture type, and result. Culture data obtained within 48 hours of hospital admission are used to set patient states and compute admission probabilities upon admission, while test results obtained at least 48 hours after admission are used to estimate model transmission parameters via particle filtering. Patient trajectories are simulated using the Gillespie Stochastic Simulation Algorithm (GSSA), tracking patients as susceptible, asymptomatically colonized, or symptomatically colonized, and an observation model is used to sample patients for testing at observed times.
Degree
Doctor of Philosophy
Major
Mathematics
Mathematics
Embargo Date
May 1, 2029