Research
and
Projects.
I study infectious disease through classic epidemiological studies, biosurveillance, and computational analysis. My current projects revolve around H5N1 highly pathogenic avian influenza (HPAI) and an independent study of early-pandemic COVID-19 cases.
HPAI Farm Biosecurity & Outbreak Risk Factors
STUDY IN DEVELOPMENTMy graduate thesis is a case-control study examining the relationship between farm-level biosecurity practices and highly pathogenic avian influenza (HPAI) outbreaks among poultry operations in the Midwestern United States. Using farm-level biosecurity and outbreak data, the study will investigate potential outbreak-associated factors through observational epidemiologic methods.
The dataset has been acquired, and data characterization, variable selection, and the analytical framework are being developed in collaboration with USDA Agricultural Research Service (ARS) scientists and Mississippi State University faculty.
USDA ARS NACA fellowship · Mississippi State University College of Veterinary Medicine · Advisor: Dr. Isaac Jumper
COVID-19 Epidemiologic Analysis
INDEPENDENT METHODOLOGICAL STUDYThis self-guided study examines the association between reported pre-existing medical conditions and hospitalization among early-pandemic COVID-19 cases, using publicly available CDC surveillance data. The project provides a pracitcal foundation in classical epidemiology, complementing my compuational infectious disease research.
The analysis progresses from crude risk ratios and risk differnces to directed acyclic graph (DAG)-informed confounder identification, multivariable logistic regression, and regression standardization.
Project Focus: Epidemiological methods · Confounding and adjustment · R and Quarto
PUBLIC GITHUB REPOSITORYOne-Month-Ahead HPAI Early-Warning Forecasting
I developed a leakage-safe forecasting framework to test whether county-level HPAI early-warning signal remains when forecasts are restricted to information strictly available at the time of prediction.
The study framework uses a 24-month rolling-origin simulation across the conterminous United States. At each forecast origin, unavailable forecast-month covariates are synthesized from historical data, and models are developed using only prior information.
The study framework evaluates four modeling families: logistic regression, tabular ensembles, recurrent-only models, and graph-temporal models. Feature analyses examine individual covariates and feature-families to see which contribute to retained predictive signal.
Forecast outputs are integrated into an interactive web dashboard, translating model outputs into interpretable biosurveillance risk tiers for state veterinarians and poultry producers.
Ongoing research · manuscript in preparation
HPAI Risk Classification & Pseudo-Forecasting
My interest in computational epidemiology research began during a USDA ARS internship, where I developed national-scale county-month risk models integrating environmental, agricultural, geospatial, and migratory-bird data. Rare-event classification achieved 75%+ balanced accuracy on 2022–2024 surveillance data with approximately 1% outbreak-positive observations.
I presented this research to USDA ARS leadership and multiple National Program Leaders. The project later received the Judge's Choice Award at UTA College of Engineering's Innovation Day 2026.
During the latter half of the internship, I examined pseudo-forecasting under temporal shift using observed future time-varying (weather-climate) covariates. This work motivated the question of whether or not early-warning signal can be retained when forecasts are restricted strictly to information available at prediction time, which my subsequent research focuses on.
Pseudo-forecasting outputs were stratified by geographical migratory flyway, season, and flyway-season, examining where and when pseudo-forecasting fails to capture outbreak signal.
Related manuscript · submitted May 2026
Blackstock, J., Maberry, W., et al. A forward-looking high pathogenicity avian influenza monthly risk model for counties in the conterminous United States. Submitted to Scientific Reports.