Supplement for Cloud Computing: Alcohol Use Disorder Treatment Simulation

PROJECT SUMMARY/ABSTRACT Marked disparities are found for alcohol use, alcohol-related problems, and utilization of alcohol health services across demographic subgroups in the United States. Advances in statistical modeling and computing capabilities have enabled the development of sophisticated simulation models helpful for

assessing potential impacts of policies and health services interventions on alcohol-related outcomes and on disparities in those outcomes. By projecting effects of different interventions across population subgroups over time, simulation modeling can help identify and prioritize specific types of alcohol health services

interventions that may best reduce disparities and promote equity in long-term recovery from alcohol use disorder (AUD). The ongoing parent grant, Alcohol Use Disorder Treatment Simulation: Modeling Treatment Impacts on Alcohol-related Disparities in the United States (R01AA029812-02), addresses the National Institute on Alcohol Abuse and Alcoholism’s goal to understand where and how to increase

access to alcohol health services across the continuum of care to promote health equity. Our simulation model is particularly well-suited for enhanced computing using cloud infrastructure. We apply the model to a geographically explicit simulated population representing the populations of California and Texas, and we

make long-term projections for AUD severity and recovery for key population subgroups over time. With supplemental funding, we would enhance the existing project by using new cloud technologies. This would facilitate more realistic simulation conditions and enable new discoveries by improving the computational

efficiencies of the research. The Specific Aims for the Supplement are to (1) build proof-of-concept examples and evaluate cloud service options capable of providing scalable cluster computing for Message Passing Interface–based simulation modeling software; (2) compare simulation results for a computationally

constrained laptop-run model with results from running the same model on a scalable cloud computing cluster; and (3) evaluate the costs for setting up and running the simulation model (including data uploads, run time, and data downloads) for execution on the cloud solutions compared to running the model on a

high-performance laptop. Receipt of supplemental funding will permit the creation of a more realistic simulation to enhance the results through the production of better estimates of health services utilization and impact. Comparisons of cloud services with non-cloud options will provide additional information to the

National Institutes of Health concerning costs associated with these specific use cases and strategies for institutions without access to traditional supercomputers to access high-performance computing through the cloud.