Date of Award

Fall 1-8-2021

Degree Type

Thesis

Degree Name

Master of Public Health (MPH)

Department

Public Health

First Advisor

Dr. Gerardo Chowell

Second Advisor

Desiree Mustaquim

Abstract

INTRODUCTION: The public health response to an emerging infectious disease epidemic is based on risk assessments that predict the severity of the health threat. Although infectious disease surveillance data is often limited in scope, mathematical models can provide meaningful information about epidemic growth dynamics to inform development of public health interventions.

AIM: This investigation aims to validate application of a sub-epidemic version of the generalized logistic-growth model (GLM) to a delineated period of epidemic growth representing the influenza season, using national surveillance data for incidence of influenza-like illness (ILI) in the United States.

METHODS: Surveillance data for ILI case counts were obtained from the Centers for Disease Control and Prevention (CDC) website, FluView. GLM models with one, two, and three sub-epidemics were fit to four epidemic growth periods across four years, each containing 30 weekly ILI incidence counts. Parameter estimates were obtained through nonlinear least squares curve-fitting and sub-epidemic curves were aggregated into the best fit model. Model performance was evaluated using calculation of performance metrics, bootstrapping, and visual analysis.

RESULTS: Model performance consistently improved across all four seasons as the number of sub-epidemics incorporated into the GLM increased (n=1 to n=3). The parameter and sub-epidemic estimates provided information about the growth dynamics of the epidemic period, identifying trends specific to each season.

DISCUSSION: The sub-epidemic GLM provides useful results about epidemic dynamics using national case count data. In addition, while logistic growth models are often applied to discrete outbreaks, the results of this investigation support application of the model to periods of epidemic growth within seasonal trends as well. The findings support the continued use of this model for academic and other public health application.

DOI

https://doi.org/10.57709/20620382

File Upload Confirmation

1

Share

COinS