Predicting air pollution changes due to temperature increases in two Brazilian capitals using machine learning - a necessary perspective for a climate resilient health future.

Abstract

Given that climate change can exacerbate the health impacts of air pollutants, we evaluated the impact of temperature increase scenarios on air pollutant levels (O₃, PM_2.5, and PM₁₀) in Porto Alegre and Recife, Brazil. Air pollutants and meteorological data were collected, and simulations were performed using a Support Vector Machine model with radial basis function kernel, applying temperature increases of 0.5°C, 1.0°C, 1.5°C, and 2.0°C to predict future pollutant concentrations. The data were analyzed seasonally and annually. Pearson correlation and principal component analyses (PCA) explored the relation with meteorological conditions. Simulations revealed that rising temperatures do not uniformly lead to increased pollutant concentrations; instead, the effects are highly dependent on local meteorological and climatic conditions. In Porto Alegre, O₃ levels increased throughout the year, with a peak of 14.14% during the summer in the + 2.0°C scenario, while PM_2.5 and PM₁₀ also showed marked seasonal increases. Conversely, in Recife, O₃ levels decreased in some seasons but increased during autumn, with particulate matter levels also rising during the summer. The findings underscore the need for health systems to consider these dynamics in their management strategies through location-specific investigations and emphasize the importance of policy-driven adaptive measures to build climate-resilient health systems.