Enhancing SPACE, an innovative python package to account for spatial confounding used to estimate climate-sensitive events among older Medicare

项目总结

Project Summary The WHO listed air pollution and climate change as two of the top ten threats in 2019, and earlier research indicates links between climate change exposures and brain health. Further, the burden of older persons with Alzheimer’s disease (AD) and related dementias (ADRD) is expected to double by 2060, with the largest increase for Hispanic Americans. The environmental impact of climate change could become a brain health emergency that we are unprepared to tackle. To date, little is known regarding impacts of heat or air pollution, including wildfire smoke, all of which are impacted by climate change, on the elderly with AD/ADRD. Our parent R01 addresses these scientific gaps by estimating the impact of both heat and air pollution on cause specific admissions, readmissions, and mortality and disseminating the statistical methods used in these analyses. Accounting for spatial confounding that results from various factors (e.g. socioeconomic, demographic, meteorological) being associated with both wildfire smoke and heat exposure and ADRD hospitalizations is critical. There are various approaches to adjust for spatial confounding, however, there are no clear guidelines on which approach should be used under which setting. To solve this problem, as part of the parent R01 we developed spacebench, a python based statistical software to compare the performance of spatial confounding algorithms using benchmark datasets representing the real data and allowing researchers to select the optimal method for a specific dataset. While spacebench is an innovative software, it was developed to prioritize functionality but more work needs to be done to ensure it follows software engineering best practices, and significant improvement is necessary to make it more accessible to a wider audience. In this administrative supplement, we propose to enhance the existing spacebench software by refactoring the existing code (Aim 1), importing the software to R by adding an R API allowing for a wider user base (Aim 2), and increasing reproducibility providing containers and documentation for cloud usage (Aim 3). The refactoring of the spacebench software package will enable a large user base of researchers to efficiently utilize the tool, add capabilities to the codebase, and offer improvements to the spatial confounding algorithms implemented in the software. Then we will apply this tool to our research on climate-related variables and AD/ADRD outcomes. The cloud readiness strategies remove specific dependencies and allow wider usability. The optimized and refactored spacebench packages will provide a superior framework for accounting for spatial confounding associated with exposure to environmental agents which will be applicable to a wide range of environmental health research.