Bayesian Statistical Methods for Air Pollution and Mental Health Research in a changing climate
The main aim of this project is to describe the long-term effect of air pollution on mental health outcomes in the UK. It is expected to be a three-year project, under PhD funding from the Medical Research Council.
- Assimilate numerical models with ground-level monitoring station and satellite-derived measurements for air pollution and climate, to create a new air pollution exposure model for the UK.
- Use machine learning algorithms to classify land use in the UK for a measure of proximity to greenspaces and typology.
- Develop a model for quantifying the effects to air pollution on mental health outcomes, while investigating if nearby greenspaces could alleviate these potential negative effects.
- Perform causal inference between long-term exposure to air pollution and mental health.
The current global focus on the climate crisis and air pollution has brought new developments in GIS methods, satellite imagery, and open source data. Similarly, the area of mental health research has rapidly gained interest from a variety of stakeholders, including international governments, research organisations, and individuals, as seen in the recent UN Change Conference. Therefore, this project positioned in the intersection between health, environment and biostatistical research fields, will provide unique insights on the relationship between polluted environment and mental health.
Underpinning this project are a number of methodological advances too. Bayesian statistics is a current significant area of interest in mathematics, especially its innate ability to incorporate multiple data sources and updated data. Similarly, the use of Big Data and machine learning are both growing fields in industry and academia. This project will take advantage of large, open-source datasets and will use machine learning for characterizing greenspaces from satellite imagery. Furthermore, causal inference is a really hot methodological topic; many current studies on air pollution and mental health just focus on correlation, even so 'correlation does not imply causation'.
Finally, by including location, physical health, and socioeconomic variables within the models, we can identify trends in mental health outcomes across the country and in different demographic and vulnerable groups. By identifying these groups, mental health awareness campaigns, treatment resources, and medical training can be distributed appropriately. Similarly, any identified trends in the effects of air pollution, climate, and greenspaces on mental health can help to form more specific targets in the battles to reduce emissions, combat climate change, and promote the development and use of greenspaces and parks for leisure and exercise.