Estimate the effect of Hurricane Florence (2018) on buprenorphine transactions for opioid use disorder in southeastern coastal North Carolina using a Bayesian generalized synthetic control approach.

Compare two perspectives on access: transactions grouped by where patients live versus where pharmacies are located.

Use streamflow data to capture flooding conditions during the hurricane.

Find disruptions in both patient- and pharmacy-level transactions, with the largest pharmacy-level decrease (about 22%) during the evacuation week.

Link to the publication.

Integrate spatially misaligned data from counties and ZIP codes to analyze the complex interactions of five opioid-related outcomes.

Apply GIS methods to align ZIP codes with ZIP Code Tabulation Areas (ZCTAs) for a more detailed exploration of the opioid epidemic, revealing critical localized impacts.

Emphasize the need for both granular and county-level data to avoid misinterpretations, particularly in rural and urban regions.

Link to the publication.

Examine trends and relationships among different outcomes believed to reflect opioid misuse.

Employ a Bayesian dynamic spatial factor model to capture the interrelated dynamics within six different county-level outcomes related to opioid misuse in North Carolina.

Investigate trends and relationships among illicit opioid overdose deaths, emergency department visits, opioid use disorder treatments, buprenorphine prescriptions, and hepatitis C and HIV cases.

Develop a novel technique within a Markov chain Monte Carlo algorithm to overcome challenges in loadings matrix estimation, enhancing model identifiability.

Provide a deeper understanding of the opioid epidemic's dynamics across time and space to inform public health strategies.

Link to the publication.

Develop a directional wind speed distribution using a Weibull distribution in such a way that the parameters of the distribution depend on wind direction.

Construct the dependence of the parameters of the Weibull distribution on wind direction using harmonic regression via weighted least squares.

Analyze the changes in wind speed and wind direction from present to future climate scenarios.

Link to the publication.

Utilize methods from extreme value theory, namely the block maxima method and peaks-over-threshold method, to investigate the potential enhancement of estimating extreme wind speeds.

Block maxima, peaks-over-thresholds, and point process methods are utilized to model the upper tail of the conditional distribution of the extreme wind speed given wind direction.

Simulation studies, analysis of output from climate model simulation, and model comparisons are discussed.

Estimate the effect of Hurricane Florence on weekly buprenorphine dispensing for patients on long-term opioid therapy across North Carolina three-digit ZIP code (ZIP3) regions.

Develop a hierarchical Bayesian synthetic control model that expresses the effect as the product of the proportion of the population displaced and the change in dispensing per unit increase in that proportion.

Use displacement estimates from the North Carolina Governor's Office to inform the prior for the displaced proportion, helping regularize estimates where the dispensing data alone carry limited information.

Develop a framework that brings together several causal inference methods to study how continuous environmental exposures relate to opioid-related health outcomes.

Part of the IMSI–NISS Ideas Lab: Data Science at the Intersection of Public Health and the Environment.

Decompose the complex structure of the spatio-temporal wind speed process into smaller components that can be estimated more easily. Then combine these components to obtain an estimate of the overall process.

A smooth space-time function is used to capture the first-order mean structure, taking into account periodicity in time, and a combination of empirical orthogonal functions (EOFs) and a first-order dynamical Gaussian process is employed to characterize the potentially complex second-order covariance structure.

A crucial aspect of the proposed model is its utilization of the annual "circularity" concept, which introduces spatio-temporal replicates allowing for flexible nonstationary space-time modeling.