Small Area Forecasting of Opioid-Related Mortality: Bayesian Spatiotemporal Dynamic Modeling Approach

BACKGROUND

Opioid-related overdose mortality has remained at crisis levels across the United States, increasing 5-fold and worsened during the COVID-19 pandemic. The ability to provide forecasts of opioid-related mortality at granular geographical and temporal scales may help guide preemptive public health responses. Current forecasting models focus on prediction on a large geographical scale, such as states or counties, lacking the spatial granularity that local public health officials desire to guide policy decisions and resource allocation.

OBJECTIVE

The overarching objective of our study was to develop Bayesian spatiotemporal dynamic models to predict opioid-related mortality counts and rates at temporally and geographically granular scales (ie, ZIP Code Tabulation Areas [ZCTAs]) for Massachusetts.

METHODS

We obtained decedent data from the Massachusetts Registry of Vital Records and Statistics for 2005 through 2019. We developed Bayesian spatiotemporal dynamic models to predict opioid-related mortality across Massachusetts' 537 ZCTAs. We evaluated the prediction performance of our models using the one-year ahead approach. We investigated the potential improvement of prediction accuracy by incorporating ZCTA-level demographic and socioeconomic determinants. We identified ZCTAs with the highest predicted opioid-related mortality in terms of rates and counts and stratified them by rural and urban areas.

RESULTS

Bayesian dynamic models with the full spatial and temporal dependency performed best. Inclusion of the ZCTA-level demographic and socioeconomic variables as predictors improved the prediction accuracy, but only in the model that did not account for the neighborhood-level spatial dependency of the ZCTAs. Predictions were better for urban areas than for rural areas, which were more sparsely populated. Using the best performing model and the Massachusetts opioid-related mortality data from 2005 through 2019, our models suggested a stabilizing pattern in opioid-related overdose mortality in 2020 and 2021 if there were no disruptive changes to the trends observed for 2005-2019.

CONCLUSIONS

Our Bayesian spatiotemporal models focused on opioid-related overdose mortality data facilitated prediction approaches that can inform preemptive public health decision-making and resource allocation. While sparse data from rural and less populated locales typically pose special challenges in small area predictions, our dynamic Bayesian models, which maximized information borrowing across geographic areas and time points, were used to provide more accurate predictions for small areas. Such approaches can be replicated in other jurisdictions and at varying temporal and geographical levels. We encourage the formation of a modeling consortium for fatal opioid-related overdose predictions, where different modeling techniques could be ensembled to inform public health policy.

Temporal Aggregation Effects in Spatiotemporal Traffic Modelling

Spatiotemporal models are a popular tool for urban traffic forecasting, and their correct specification is a challenging task. Temporal aggregation of traffic sensor data series is a critical component of model specification, which determines the spatial structure and affects models’ forecasting accuracy. Through extensive experiments with real-world data, we investigated the effects of the selected temporal aggregation level for forecasting performance of different spatiotemporal model specifications. A set of analysed models include travel-time-based and correlation-based spatially restricted vector autoregressive models, compared to classical univariate and multivariate time series models. Research experiments are executed in several dimensions: temporal aggregation levels, forecasting horizons (one-step and multi-step forecasts), spatial complexity (sequential and complex spatial structures), the spatial restriction approach (unrestricted, travel-time-based and correlation-based), and series transformation (original and detrended traffic volumes). The obtained results demonstrate the crucial role of the temporal aggregation level for identification of the spatiotemporal traffic flow structure and selection of the best model specification. We conclude that the common research practice of an arbitrary selection of the temporal aggregation level could lead to incorrect conclusions on optimal model specification. Thus, we recommend extending the traffic forecasting methodology by validation of existing and newly developed model specifications for different temporal aggregation levels. Additionally, we provide empirical results on the selection of the optimal temporal aggregation level for the discussed spatiotemporal models for different forecasting horizons.

Chronic wasting disease undermines efforts to control the spread of brucellosis in the Greater Yellowstone Ecosystem

Wildlife diseases pose a substantial threat to the provisioning of ecosystem services. We use a novel modeling approach to study the potential loss of these services through the imminent introduction of chronic wasting disease (CWD) to elk populations in the Greater Yellowstone Ecosystem (GYE). A specific concern is that concentrating elk at feedgrounds may exacerbate the spread of CWD, whereas eliminating feedgrounds may increase the number of elk on private ranchlands and the transmission of a second disease, brucellosis, from elk to cattle. To evaluate the consequences of management strategies given the threat of two concurrent wildlife diseases, we develop a spatiotemporal bioeconomic model. GPS data from elk and landscape attributes are used to predict migratory behavior and population densities with and without supplementary feeding. We use a 4,800 km² area around Pinedale, Wyoming containing four existing feedgrounds as a case study. For this area, we simulate welfare estimates under a variety of management strategies. Our results indicate that continuing to feed elk could result in substantial welfare losses for the case-study region. Therefore, to maximize the present value of economic net benefits generated by the local elk population upon CWD's arrival in the region, wildlife managers may wish to consider discontinuing elk feedgrounds while simultaneously developing new methods to mitigate the financial impact to ranchers of possible brucellosis transmission to livestock. More generally, our methods can be used to weigh the costs and benefits of human-wildlife interactions in the presence of multiple disease risks.

A statistical forecasting approach to metapopulation viability analysis

Conservation of at-risk species is aided by reliable forecasts of the consequences of environmental change and management actions on population viability. Forecasts from conventional population viability analysis (PVA) are made using a two-step procedure in which parameters are estimated, or elicited from expert opinion, and then plugged into a stochastic population model without accounting for parameter uncertainty. Recently developed statistical PVAs differ because forecasts are made conditional on models fitted to empirical data. The statistical forecasting approach allows for uncertainty about parameters, but it has rarely been applied in metapopulation contexts where spatially explicit inference is needed about colonization and extinction dynamics and other forms of stochasticity that influence metapopulation viability. We conducted a statistical metapopulation viability analysis (MPVA) using 11 yr of data on the federally threatened Chiricahua leopard frog (Lithobates chiricahuensis) to forecast responses to landscape heterogeneity, drought, environmental stochasticity, and management. We evaluated several future environmental scenarios and pond restoration options designed to reduce extinction risk. Forecasts over a 50-yr time horizon indicated that metapopulation extinction risk was <4% for all scenarios, but uncertainty was high. Without pond restoration, extinction risk is forecasted to be 3.9% (95% CI 0-37%) by year 2066. Restoring six ponds by increasing their hydroperiod reduced extinction risk to <1% and greatly reduced uncertainty (95% CI 0-2%). Our results suggest that managers can mitigate the impacts of drought and environmental stochasticity on metapopulation viability by maintaining ponds that hold water throughout the year and keeping them free of invasive predators. Our study illustrates the utility of the spatially explicit statistical forecasting approach to MPVA in conservation planning efforts.

Deforestation Increases the Risk of Scrub Typhus in Korea

Background: Scrub typhus is an important public health issue in Korea. Risk factors for scrub typhus include both individual-level factors and environmental drivers, and some are related to the increased density of vector mites and rodents, the natural hosts of the mites. In this regard, deforestation is a potential risk factor, because the deforestation-induced secondary growth of scrub vegetation may increase the densities of mites and rodents. To examine this hypothesis, this study investigated the association between scrub typhus and deforestation. Methods: We acquired district-level data for 2006-2017, including the number of cases of scrub typhus reported annually, deforestation level, and other covariates. Deforestation was assessed using preprocessed remote-sensing satellite data. Bayesian regression models, including Poisson, negative binomial, zero-inflated Poisson, and zero-inflated negative binomial models, were examined, and spatial autocorrelation was considered in hierarchical models. A sensitivity analysis was conducted using different accumulation periods for the deforestation level to examine the robustness of the association. Results: The final models showed a significant association between deforestation and the incidence of scrub typhus (relative risk = 1.20, 95% credible interval = 1.15-1.24). The sensitivity analysis gave consistent results, and a potential long-term effect of deforestation for up to 5 years was shown. Conclusion: The results support the potential public health benefits of forest conservation by suppressing the risk of scrub typhus, implying the need for strong engagement of public health sectors in conservation issues from a One Health perspective.

Black swans in space: modeling spatiotemporal processes with extremes

In ecological systems, extremes can happen in time, such as population crashes, or in space, such as rapid range contractions. However, current methods for joint inference about temporal and spatial dynamics (e.g., spatiotemporal modeling with Gaussian random fields) may perform poorly when underlying processes include extreme events. Here we introduce a model that allows for extremes to occur simultaneously in time and space. Our model is a Bayesian predictive-process GLMM (generalized linear mixed-effects model) that uses a multivariate-t distribution to describe spatial random effects. The approach is easily implemented with our flexible R package glmmfields. First, using simulated data, we demonstrate the ability to recapture spatiotemporal extremes, and explore the consequences of fitting models that ignore such extremes. Second, we predict tree mortality from mountain pine beetle (Dendroctonus ponderosae) outbreaks in the U.S. Pacific Northwest over the last 16 yr. We show that our approach provides more accurate and precise predictions compared to traditional spatiotemporal models when extremes are present. Our R package makes these models accessible to a wide range of ecologists and scientists in other disciplines interested in fitting spatiotemporal GLMMs, with and without extremes.

Extending Participatory Sensing to Personal Exposure Using Microscopic Land Use Regression Models

Personal exposure is sensitive to the personal features and behavior of the individual, and including interpersonal variability will improve the health and quality of life evaluations. Participatory sensing assesses the spatial and temporal variability of environmental indicators and is used to quantify this interpersonal variability. Transferring the participatory sensing information to a specific study population is a basic requirement for epidemiological studies in the near future. We propose a methodology to reduce the void between participatory sensing and health research. Instantaneous microscopic land-use regression modeling (µLUR) is an innovative approach. Data science techniques extract the activity-specific and route-sensitive spatiotemporal variability from the data. A data workflow to prepare and apply µLUR models to any mobile population is presented. The µLUR technique and data workflow are illustrated with models for exposure to traffic related Black Carbon. The example µLURs are available for three micro-environments; bicycle, in-vehicle, and indoor. Instantaneous noise assessments supply instantaneous traffic information to the µLURs. The activity specific models are combined into an instantaneous personal exposure model for Black Carbon. An independent external validation reached a correlation of 0.65. The µLURs can be applied to simulated behavioral patterns of individuals in epidemiological cohorts for advanced health and policy research.

Genome-scale modelling of microbial metabolism with temporal and spatial resolution

Most natural microbial systems have evolved to function in environments with temporal and spatial variations. A major limitation to understanding such complex systems is the lack of mathematical modelling frameworks that connect the genomes of individual species and temporal and spatial variations in the environment to system behaviour. The goal of this review is to introduce the emerging field of spatiotemporal metabolic modelling based on genome-scale reconstructions of microbial metabolism. The extension of flux balance analysis (FBA) to account for both temporal and spatial variations in the environment is termed spatiotemporal FBA (SFBA). Following a brief overview of FBA and its established dynamic extension, the SFBA problem is introduced and recent progress is described. Three case studies are reviewed to illustrate the current state-of-the-art and possible future research directions are outlined. The author posits that SFBA is the next frontier for microbial metabolic modelling and a rapid increase in methods development and system applications is anticipated.

Efficient and automatic methods for flexible regression on spatiotemporal data, with applications to groundwater monitoring

Fitting statistical models to spatiotemporal data requires finding the right balance between imposing smoothness and following the data. In the context of P-splines, we propose a Bayesian framework for choosing the smoothing parameter, which allows the construction of fully automatic data-driven methods for fitting flexible models to spatiotemporal data. An implementation, which is highly computationally efficient and exploits the sparsity of the design and penalty matrices, is proposed. The findings are illustrated using a simulation study and two examples, all concerned with the modelling of contaminants in groundwater. This suggests that the proposed strategy is more stable that competing methods based on the use of criteria such as generalised cross-validation and Akaike's Information Criterion. © 2015 The Authors. Environmetrics Published by John Wiley Sons Ltd.