Biostatistics seminar
This event is titled "From Map to Action: Leveraging Bayesian spatiotemporal models for targeted public health surveillance" and will feature Assistant Professor Joanne Kim.
About
Joanne Kim is an assistant professor in the biomedical informatics department.
Abstract
Bayesian spatiotemporal modeling has become an essential component of modern public health surveillance, offering a robust framework to manage complex, multi-dimensional data. This talk will discuss public health surveillance efforts using a Bayesian spatiotemporal framework inspired by two distinct public health challenges: the COVID-19 pandemic and the opioid overdose death crisis.
The first portion of the talk introduces a novel Bayesian surveillance metric designed to detect emerging, elevated-risk clusters during infectious disease outbreaks. This proposed metric integrates three core components—an area’s intrinsic risk profile, its temporal risk trend, and spatial neighborhood influence. To balance these dimensions, we introduce a weighting scheme that accommodates the specific transmission characteristics and spatial trends of an outbreak. The second portion focuses on the development of predictive spatiotemporal models for opioid overdose deaths in Ohio. By evaluating models that incorporate various data—Urine Drug Test (UDT) results, Emergency Medical Services (EMS) records, and socioeconomic factors—we demonstrate that UDT data helps improve the prediction of overdose mortality, especially during the COVID-19 pandemic. Together, these two studies showcase how advanced geospatial modeling translates complex data into actionable insights, enhancing active surveillance efforts and guiding targeted public health interventions.
Contact
Andy Ni