Date of Award

Fall 11-18-2022

Embargo Period

11-25-2022

Document Type

Dissertation - MUSC Only

Degree Name

Doctor of Philosophy (PhD)

Department

Public Health Sciences

Abstract

The COVID-19 Pandemic for the last three years brought enormous challenges to public health surveillance and underscored the importance of developing and maintaining robust systems for accurate surveillance. Keeping pace with public health data collection efforts, there is a critical need for infectious disease modeling researchers to continue to develop surveillance metrics and statistical models to accurately predict future disease trends and high-risk regions.
In the first aim, we evaluated the performance of Bayesian spatio-temporal models for the infectious disease outbreak whether these models accommodate large variability of the data and different historic data usage for timely prediction. Our choice of the likelihood and spatio-temporal mean models was influenced by the length of the past data and the variability of the data. Through the evaluation of these models, we provided future infectious disease outbreak modeling guidelines for Bayesian spatio-temporal analysis.
In the second aim, the novel cluster prediction surveillance metric based on a Bayesian spatio-temporal model was proposed. Exceedance probability, which has been commonly used for hotspot detection in statistical epidemiology, was extended to predict high-risk clusters. The proposed metric consists of three components: the area's own risk profile, temporal risk trend, and spatial neighborhood influence. We also introduced a weighting scheme to balance these three components, which accommodates the characteristics of the infectious disease outbreak, spatial properties, and disease trends. Through simulation study and real data analysis, the optimal weighting schemes were studied, and the performance of the proposed cluster prediction surveillance metric was evaluated.
Even though the COVID-19 pandemic drove the rapid growth of public health informatics area, the visualization tools specialized in neighborhood investigation and disease hotspots/clusters of spatio-temporal domains are limited. In the third aim, the software of the neighborhood clustering surveillance system is introduced. It was developed with R Shiny technology and provides an interactive map and time-series plotting tool. The county-level United States map on both entire state-level disease mapping and subsequent in-depth neighborhood analyses was provided. This visualization software will contribute to spatial epidemiology research and public health surveillance to provide an easy-to-use and effective graphics tool and facilitate the understanding of public health practitioners and public audiences.

Rights

Copyright is held by the author. All rights reserved.

Download

Share

COinS