Research

The research focuses on identifying regional disparities in public health crisis response and understanding the structural and institutional factors that shape these disparities. Various analytical methodologies, including causal inference, machine learning, geographic information systems (GIS), and natural language processing, are utilized to explore these disparities. The research is framed around three key perspectives: Public Health Crisis Management, Public Health Services Planning, and the Social Impact of Public Health Crises.

1. Public Health Crisis Management

Opioid overdose mortality is a central area of current research. A spatial regression discontinuity analysis examined whether opioid involvement in drug-related deaths changes discontinuously at the boundary between the City of Chicago and suburban Cook County, using mortality microdata from the Cook County Medical Examiner’s Office (published in Spatial and Spatio-temporal Epidemiology). The results showed that opioid involvement was consistently higher on the Chicago side of the boundary, indicating that institutional capacity and service infrastructure can produce localized discontinuities in mortality risk even when environmental conditions remain spatially continuous.

An earlier collaborative study in Nigeria, funded by the United States Agency for International Development, identified tuberculosis hotspots using geospatial early warning methods (published in JMIR Public Health and Surveillance). This study demonstrated that active case-finding in areas identified as high-risk led to higher diagnostic yield than in non-hotspot areas, establishing the value of spatial early warning systems for resource-constrained crisis response.

2. Public Health Services Planning

Research in this area focuses on demand forecasting and predictive resource allocation. Foundational work in this area developed a spatio-temporal graph convolutional network to forecast emergency medical services (EMS) demand in Busan, South Korea, incorporating regional connectivity and temporal variation into resource planning (Dissertation Chapter 2). Building on this methodology, a follow-up study developed a unified spatio-temporal graph convolutional network framework to forecast opioid-involved mortality across five heterogeneous U.S. regions using harmonized grid-based spatial units and a shared adjacency structure (AMIA Annual Symposium Proceedings). This opioid surveillance framework demonstrated that a model trained in one jurisdiction can generalize to other jurisdictions with distinct spatial configurations, supporting the development of transferable forecasting tools for multi-jurisdictional crisis response.

Earlier research on traffic accident patterns and emergency response times in Nigeria, funded by the National Institutes of Health, revealed regional disparities in emergency service accessibility and proposed strategies for improving response efficiency (published in BMC Public Health).

4. Ongoing Paper and Future Research Directions

1) Community-Based Health Disparities

Future research will explore health disparities at the community and institutional level, focusing on the impact of regional characteristics, infrastructure, and access to resources on health outcomes. Projects in this area include:

Disaster Shelter Accessibility: An ongoing study represents hospital-to-hospital patient transfers in Florida as a spatial network to examine how bed capacity is redistributed across regions during disaster and surge conditions. The research aims to identify structural vulnerabilities in inter-hospital coordination and inform more resilient regional response planning.
Machine Learning Approaches to Regional Health Policy Disparities: A planned line of research will apply machine learning methods to examine how regional disparities in healthcare resource allocation emerge and persist across jurisdictions. Building on the spatial and predictive modeling approaches developed in the opioid and emergency medical services research, this work will investigate whether machine learning–based risk detection can reveal disparities in resource distribution that are not captured by conventional administrative reporting. The goal is to identify which regional characteristics most consistently predict under-resourced areas and to assess whether these patterns hold across different health policy domains rather than being specific to a single crisis type.

2) Extending Machine Learning to Policy Analysis in Public Health

This area extends predictive machine learning methodology into the domain of policy analysis, focusing on how spatiotemporal prediction can inform comparative policy evaluation across jurisdictions. Current and planned projects include:

Cross-Jurisdictional Generalization of Predictive Frameworks: Building on the unified spatio-temporal graph convolutional network framework developed for opioid mortality forecasting, future research will examine the extent to which predictive structures learned in one policy jurisdiction transfer to jurisdictions with different governance structures, resource environments, and population characteristics. This line of work treats generalizability itself as a policy-relevant question, since a forecasting tool that fails to transfer across administrative boundaries offers limited value for multi-jurisdictional policy planning. The goal is to identify which structural features of a jurisdiction determine whether a shared predictive model can be applied without retraining, and to translate this into practical guidance for jurisdictions considering the adoption of predictive surveillance tools.
Polysubstance Risk Patterns in Opioid Overdose: A related study under review examines co-involvement patterns between opioids and emerging sedative substances, extending the surveillance framework beyond single-substance risk detection toward more granular characterization of overdose risk.

3) Spatial Analysis and Geographical Accessibility

Spatial analysis continues to be central to understanding how geographic and institutional factors shape healthcare access and resource allocation. Future research will extend the boundary-focused approach used in the Chicago opioid study to other public health contexts, examining how jurisdictional and administrative structures shape resource accessibility across different types of public health crises.

Publications

[8] DOHYO JEONG, Delcher, C., & Harris, D. (2026). Cold Exposure and Urban Opioid Risk: A Spatial Regression Discontinuity Analysis in Chicago. Spatial and Spatio-temporal Epidemiology, 100812. https://doi.org/10.1016/j.sste.2026.100812

Research Topic: Environmental exposure impacts on opioid-related overdose deaths.
Design/Method: Spatial Regression Discontinuity Design.

[7] Ku, M., DOHYO JEONG, Lee, K. H., & Choi, S. (2026). A Spatial-Network Framework for Load Balancing and Patient Transfer Management During Emergencies. Disaster Medicine and Public Health Preparedness, 20, e123. https://doi.org/ 10.1017/dmp.2026.10391

Research Topic: Patient Transfer Network across Hospital Emergency Trauma Centers
Design/Method: Spatial Network Analysis.

[6] DOHYO JEONG, & Kim, D. (2025). Bridging Gaps or Widening Disparities? A Spatiotemporal Analysis of COVID‐19 Therapeutic Distribution Policies in Texas. The International Journal of Health Planning and Management. https://doi.org/10.1002/hpm.70045

Research Topic: Identifying medical resource supply and demand imbalance and determining factors
Design/Method: Spatial-Temporal Integrated Nested Laplace Approximations (INLA).

[5] DOHYO JEONG, Jessi Hanson-DeFusco, and Dohyeong Kim. (2022). Digital Mass Hysteria during Pandemics: A Case Study of Twitter Communication Patterns in the US during COVID-19 Period. Behavioral Sciences, 14(5), 389. https://doi.org/10.3390/bs14050389

Research Topic: Analysis of changes in public sentiment regarding vaccine supply policy
Design/Method: Text Sentiment Analysis and Interrupted Time Series Analysis.

[4] Hong, S., DOHYO JEONG, & Kim, P. (2024). Have offender demographics changed since the COVID-19 Pandemic? Evidence from money mules in South Korea. Journal of Criminal Justice, 91, 102156. https://doi.org/10.1016/j.jcrimjus.2024.102156 (IF: 5.5)

Research Topic: Exploring the Influence of the COVID-19 Pandemic on Crime Patterns
Design/Method: Interrupted Time Series Analysis.

[3] DOHYO JEONG., Kim, D., Mohiuddin, H., Kang, S., & Kim, S. (2023). Regional Disparity in the Educational Impact of COVID-19: A Spatial Difference-in-Difference Approach. Sustainability, 15(16), 12514. https://doi.org/10.3390/su151612514 (IF: 3.9)

Research Topic: Analyzing regional disparities in transitioning to online classes.
Design/Method: Spatial Difference-in-Difference

[2] Odusola, A. O., DOHYO JEONG, Malolan, C., Kim, D., Venkatraman, C., Kola-Korolo, O., … & Nwariaku, F. E. (2023). Spatial and temporal analysis of road traffic crashes and ambulance responses in Lagos state, Nigeria. BMC public health, 23(1), 2273. https://doi.org/10.1186/s12889-023-16996-8 (IF: 4.7)

Research Topic: Comparing traffic accident patterns based on time, and regional characteristics.
Design/Method: Geospatial mapping and Hotspot Analysis.

[1] Ogbudebe Chidubem, DOHYO JEONG, OdumeBethrand ….. (2023). Editorial Decision/Comments on “Identifying Hotspots of Tuberculosis in Nigeria using Early Warning Outbreak Recognition System: Retrospective Analysis of Implications for Active Case Finding Interventions. JMIR Public Health and Surveillance, 9 (1), e40311. https://publichealth.jmir.org/2023/1/e40311/ (IF: 8.5)

Research Topic: Evaluation of the effectiveness of tuberculosis early warning program
Design/Method: Kernel density and Getis-Ord Gi* Hot Spot Analyses.

Publications in Korea

[5] CHANG-JIN KIM, DOHYO JEONG, (2022). Factors Influencing Public Officials Innovative Behavior for Platform Governance. The Journal of Korea Policy Research. Vol.22 No.3: 141-171.

Research Topic: Analyzing innovation factors in public sector platform governance.
Design/Method: Platform governance and Factors influencing innovative behavior.

[4] DOHYO JEONG, Sangho Moon, SUHO BAE. (2019). Factors Affecting the Distribution of National Subsidies in Korean Local Governments: Focusing on Rhodes’ Power-Dependence Model. The Korea Journal of Policy Analysis and Management, Vol.29 No3: 21-53.

Research Topic: Factors in central subsidy allocation to local governments
Design/Method: Panel Corrected Standard Errors (PCSE) model and Prais-Winsten procedure.

[3] DOHYO JEONG, CHANG-JIN KIM, SUHO BAE. (2019). A Study on Determinants of Tax Attitude: Focusing on Slippery Slope Framework, Public Policy Review, Vol.33 No.3: 43-72.

Research Topic: Examining the Impact of Tax Compliance and Taxpayer Attitude
Design/Method: Decision tree model and neural network model.

[2] DOHYO JEONG, YOUNGKYU LEE, SEONGYOUNG JEONG. (2018). An Analysis of the Effect of the Tax Rate on the Financial Efficiency of Local Governments. The Korea Journal of Local Government Studies, Vol.22 No.3: 415-443.

Research Topic: Local Tax Flexibility and Fiscal Efficiency
Design/Method: Difference-in-Difference analysis, Panel Corrected Standard Errors model.

[1] Dae-yong Hyun, DOHYO JEONG, (2017). Analysis of differences in perception of administrative values among civil servants and general civil servants: Focused on Suwon City Government Officials. Suwon Research Institute. No. 12: 119-141.

Research Topic: Comparing administrative values between City officials and general civil servants.
Design/Method: Ordered Logit Model.

Ongoing Paper

[4] DOHYO JEONG, Chris Delcher, Nick Anthonya, Daniel Harris, Daniel Harris. (2026). Neighborhood-Level Patterns in Bromazolam-involved Polysubstance Overdose Mortality.

Research Topic: Sociodemographic and Regional Effects on Bromazolam-Related Overdose Incidence Patterns.
Design/Method: Spatial Integrated Nested Laplace Approximations (INLA).

[3] DOHYO JEONG, Dohyeong Kim, Sunghwan Cho. (2026). Where and When? Predicting Emergency Medical Service Demands through Spatial-Temporal Graph Convolutional Networks.

Research Topic: Identification of medical resource imbalances and optimal allocation.
Design/Method: ST-Graph Convolutional Network and Maximal Covering Location Problem.

[2] DOHYO JEONG, Dohyeong Kim, Chang-jin Kim. (2026). Fiscal Forecast Errors in Public Health Expenditure: Based on a Spatial-Temporal Approach.

Research Topic: Identifying medical resource supply and demand imbalance and determining factors
Design/Method: Calculating Regional Variability, Spatial Regression Model.

[1] DOHYO JEONG, Dohyeong Kim, Okey Okuzu, Chidubem Ogbudebe. (2026). Beyond Distance: Does Geographical Accessibility for Tuberculosis Treatment Follow Supply, Demand or Both?

Research Topic: The impact of geographical accessibility on tuberculosis treatment initiation
Design/Method: Calculating the regional geographic accessibility, Generalized Linear Mixed Models

Conference Presentation

[6] DOHYO JEONG. (2025). Where and When? Predicting Emergency Medical Service Demands through Spatial-Temporal Graph Convolutional Networks. ASPA (American Society for Public Administration) 2025 Annual Conference. March. 2025.

[5] DOHYO JEONG. (2024). Beyond Distance: Does Geographical Accessibility for Tuberculosis Treatment Follow Supply, Demand or Both? APHA (American Public Health Association) 2024 Annual Conference. Nov. 2024.

[4] DOHYO JEONG. (2023). Differential Side Effects of COVID-19 Response Policies on the U.S. Labor Market: A Spatial-Temporal Analysis. APPAM (Association for Public Policy Analysis and Management) 2023 Annual Conference. Nov. 2023.

[3] DOHYO JEONG. (2023). The Patterns of COVID-19 Therapeutics Supply and Demand in Texas: A Spatial-Temporal INLA Approach. APHA (American Public Health Association) 2023 Annual Conference. Nov. 2023.

[2] DOHYO JEONG. (2023). Regional disparity in the uninsurance rate impact of COVID-19: a spatial machine learning approach. ASPA (American Society for Public Administration) 2023 Annual Conference. 21. March. 2023.

[1] DOHYO JEONG. (2023). Digital Mass Hysteria? during Pandemics: A Case Study of Twitter Communication Patterns in the US during COVID-19 Period. Conference On Public Process Research. 12. Jan. 2023.

Conference Presentation in Korea

[7] DOHYO JEONG. (2022). A comparison of the spread trend prediction model according to the government’s COVID-19 response policy change and its influence, 2022 Korean Public Administration International Conference. Korea. 22 June. 2022.

[6] DOHYO JEONG. (2021). The effect of the government’s vaccination management plan on the change of sentiment toward vaccines, 2021 Global Disastronomy Workshop. Texas. USA. 17 Dec. 2021.

[5] DOHYO JEONG, Chang-jin Kim. SUHO BAE. (2019). A Study on the Factors Affecting the Taxation Attitude of General Taxpayers. Korean Association for Local Government Studies Winter Conference. Seoul. KOREA. 14 Feb. 2019

[4] Chang-jin Kim. DOHYO JEONG, SUHO BAE. (2019). The Mediation Effect of Dispute Settlement System in the Perception of Regional Dispersion and Intergovernmental Relations. Korean Association for Local Government Studies Winter Conference. Seoul. KOREA. 14 Feb. 2019.

[3] DOHYO JEONG, YOUNGKYU LEE, SUHO BAE. (2018) An Analysis of the Effect of the Tax Rate on the Financial Efficiency of Local Governments. Korea Association of Local Administration Summer Joint Conference Chungcheong-do. KOREA. 20 Jul. 2018

[2] DOHYO JEONG. (2017). The Effects of Tax Recognition on the pros and cons of Welfare Policy. Seoul Association of Public Administration Fall Conference. Seoul. KOREA. 3 Nov. 2017

[1] Dae-yong Hyun, DOHYO JEONG. (2017). A Study on the Policy Diffusion of Local Government in Korea: focusing on Resident Participation Budget System. Korea Association of Local Administration Summer Joint Conference. Gyeonggi-do. KOREA. 18 Aug. 2017