Research

My dissertation treats opioid-epidemic response as a connected decision process. Spatial and statistical models identify mismatches among opioid-related burden, harm-reduction resources, and geographic access. Stochastic bilevel optimization examines how illicit supply networks may reroute or substitute supply after disruption, while agent-based modeling evaluates how interventions and changing supply conditions affect individuals and communities over time. The framework below illustrates how these studies inform one another.

Looking ahead, I will extend this work from identifying and evaluating problems to recommending adaptive intervention strategies. My future research will develop methods for designing intervention portfolios under resource constraints, anticipating behavioral and network responses that may shift harm, and updating intervention plans as new evidence becomes available. This agenda will begin in substance-use response and access-to-care systems, with extensions to other public-sector service systems.

Integrated Opioid-Epidemic Research Framework

Spatial Analysis of Naloxone Access and Opioid-Related Outcomes

Stochastic Optimization of Fentanyl Supply Disruption and Harm Mitigation

Agent-Based Modeling of Opioid Use and Harm-Reduction Interventions

Additional Operations Research Applications

Mixed-Model Sequencing and Metaheuristics for Assembly Systems

Methods & Computational Tools

Integrated Opioid-Epidemic Research Framework

Framework connecting spatial analysis of access gaps and stochastic optimization of supply adaptation with agent-based modeling of intervention performance and its dynamic and distributional consequences.

Spatial Analysis of Naloxone Access and Opioid-Related Outcomes

Map of South Carolina classifying communities by opioid-related burden and naloxone access: high burden and low access, low burden and high access, high burden and high access, low burden and low access, and other or unclassified areas.

Burden-and-access typology for South Carolina communities. The map classifies areas according to their relative opioid-related burden and access to naloxone resources.

This research examines how naloxone access and opioid-related health outcomes vary across South Carolina—and how the patterns change across geographic scales. At the county level, I use spatial regression to examine associations among naloxone availability, overdose mortality, and community characteristics. At the ZIP level, I use negative-binomial regression to evaluate relationships among opioid-related emergency-department utilization, proximity to naloxone distribution sites, healthcare access, and socioeconomic conditions.

To translate these analyses into planning-relevant evidence, I developed a burden-and-access typology that classifies communities according to their relative opioid-related burden and harm-reduction access. Initial findings indicate that the alignment between naloxone access and opioid-related burden differs by geographic scale and that proximity-based measures can reveal access patterns not captured by simple resource counts. Together, these analyses help identify communities experiencing both elevated burden and limited access and support geographically targeted harm-reduction strategies.

Methods: Spatial regression, negative-binomial regression, GIS, geographic accessibility analysis, and burden-and-access typology development

Selected outputs: INFORMS Annual Meeting 2024; IISE Annual Conference & Expo 2025

Manuscripts under review:

Jahan Beikloo, M., Kurz, M. E., Toyoda, M., Suiter, E., Miller, B. L., Kang, H. “Regression Analysis of Naloxone Availability, Opioid Overdose Mortality, and ED Visits in South Carolina.”
Toyoda, M., Kurz, M. E., Miller, B. L., Jahan Beikloo, M., Kang, H. “Beyond the Overdose: Examining Naloxone Administration and Repeated Emergency Department Visits.”

Stochastic Optimization of Fentanyl Supply Disruption and Harm Mitigation

This research develops a scenario-based stochastic bilevel optimization framework for coordinating fentanyl supply-disruption and harm-reduction strategies under uncertainty. An upper-level decision-maker selects interventions, while an adaptive illicit supply network responds by rerouting flow and substituting among supply sources. Scenarios represent uncertainty in fentanyl potency and the resulting public-health harm.

The model evaluates how disrupting established supply pathways may redirect supply toward backup or less familiar sources and alter the distribution of harm. It compares intervention strategies using expected-harm and tail-risk objectives and examines how network structure, supply stability, potency uncertainty, and harm-reduction reach affect outcomes. The goal is to identify when coordinated strategies reduce harm—and when network adaptation may produce limited benefits or unintended public-health consequences.

Diagram of a three-tier fentanyl supply network. Two wholesale suppliers connect to three mid-level distributors, which supply five local distributors through multiple and partially overlapping pathways, illustrating opportunities for supply rerouting after disruption.

Illustrative three-tier fentanyl supply network represented in the stochastic bilevel optimization model.

Methods: Stochastic programming, bilevel optimization, scenario-based modeling, network optimization, risk-sensitive decision-making, and computational experimentation

Manuscript in prepration: Jahan Beikloo, M., Sharkey, T., Kurz, M. E., Janan, F., “Coordinated Harm Mitigation in Illicit Fentanyl Supply Network Interdiction Under Uncertainty: A Stochastic Bilevel Approach.”

Diagram of Oconee County containing red points that represent simulated agents, blue triangles for treatment locations, orange squares for naloxone resources, and green circles for healthcare locations. Surrounding illustrations represent naloxone, medications, emergency medical response, and community healthcare.

Illustrative spatial initialization of the Oconee County agent-based model, showing simulated agents and the locations of treatment, naloxone, and healthcare resources.

Agent-Based Modeling of Opioid Use and Harm-Reduction Interventions

This research develops a county-level agent-based model of opioid use, illicit supply dynamics, overdose risk, and community interventions, using Oconee County, South Carolina, as a pilot setting. The model represents heterogeneous individuals whose opioid-use and health states evolve through behavioral, treatment, healthcare, and justice-system interactions. It connects individual behavior with supply conditions by representing drug purchasing, primary and backup dealer relationships, dealer inventory and restocking, variation in drug potency, and fatal and nonfatal overdose risk.

The model also represents access to naloxone, treatment, emergency medical response, peer support, law enforcement, and other community resources. It provides a simulation testbed for evaluating how interventions perform over time and how their effects vary across individuals, locations, and supply conditions. Planned experiments will compare changes in treatment access, harm-reduction availability, intervention timing, and supply-related risk, with particular attention to geographic and distributional differences in outcomes.

Methods: Agent-based modeling, geospatial initialization, empirical calibration, sensitivity analysis, scenario design, and policy experimentation

Selected presentations: INFORMS Annual Meeting 2025; Southern Criminal Justice Association Conference 2024 (related early-stage work)

Manuscript in prepration: Jahan Beikloo, M., Kurz, M. E., Suiter, E., Miller, B. L., “Simulating Opioid Use, Overdose Risk, and Harm-Reduction Interventions: A County-Level Agent-Based Model.”

State-transition diagram with eight states: no use, medical use, recreational use, heavy use, opioid use disorder, treatment, opioid-related overdose, and death. Blue arrows represent behavioral, health, and treatment transitions; purple arrows represent overdose-related transitions; and a dashed blue pathway represents background mortality from any living state.

Opioid-use, treatment, overdose, and mortality states represented in the agent-based model.

Additional Operations Research Applications

Mixed-Model Sequencing and Metaheuristics for Assembly Systems

Alongside my public-health research, I collaborate on optimization research for mixed-model automotive assembly systems. This work examines sequencing decisions designed to reduce work overload while representing operational features such as interacting stations, open stations, and task-skipping policies.

I contribute to the development and computational evaluation of a customized Greedy Randomized Adaptive Search Procedure (GRASP) using industrial-scale problem instances. My work includes Python implementation, computational experiment design, comparisons of solution quality and runtime, and validation against integer-programming benchmarks. Although the application differs from my public-sector research, it extends my experience in scalable optimization, algorithm design, and computational experimentation.

Methods: Mixed-model sequencing, combinatorial optimization, GRASP, integer programming, and computational benchmarking

Manuscript in revision: Alghazi, A., Kurz, M. E., Jahan Beikloo, M., Al Hanbali, A. “The Mixed Model Sequencing Problem with Skip Policy and Open Stations.”, Computers & Industrial Engineering.

Methods & Computational Tools

Optimization & Decision Modeling

Stochastic programming, bilevel optimization, mathematical and network optimization, scenario-based analysis, and risk-aware decision-making

Simulation & Computational Experimentation

Agent-based modeling, simulation-optimization, experimental design, sensitivity analysis, empirical calibration, and policy experimentation

Statistical & Spatial Analytics

Spatial regression, count-data modeling, GIS, geographic accessibility analysis, data visualization, and community typology development

Software & Computing

Python, R, Gurobi, ArcGIS, Git/GitHub, and high-performance computing environments

Page updated

Google Sites

Report abuse