Land surface temperature predicts mortality due to chronic obstructive pulmonary disease: a study based on climate variables and impact machine learning

Chronic Obstructive Pulmonary Disease (COPD) has been the focus of scientists and policymakers in the past decade with regard to mortality rates and global warming. The long-term shift in temperature and weather patterns, commonly called climate change, is an important public health issue, especially concerning COPD. Using the most recent county-level age-adjusted COPD mortality rates among adults older than 25 years, this study aimed to investigate the spatial trajectory of COPD in the United States between 2001 and 2020. Global Moran’s I was used to investigate spatial relationships utilising data from Terra satellite for night-time Land Surface Temperatures (LST_nt), which served as an indicator of warming within the same time period across the United States. The Forest-based Classification and Regression model (FCR) was applied to predict mortality rates. It was found that COPD mortality over the study period was spatially clustered in certain counties. Moran’s I statistic (0.18) showed that the COPD mortality rates increased with LST_nt, with the strongest spatial association in the eastern and south-eastern counties. The FCR model successfully predicted mortality rates in the study area using LST_nt values, achieving an R² value of 0.68, which accounted for COPD mortality rates independently. Policymakers in the United States could use the findings of this study to develop long-term spatial and health-related strategies to reduce the vulnerability to global warming of patients with acute respiratory symptoms.