Google Maps LinkNEW ZOOM Link!!!: https://lu-se.zoom.us/j/68139546348
Air pollution is a global environmental risk and a leading cause of poor health, responsible for one in every eight deaths worldwide according to the World Health Organization (WHO). In Africa, the lack of air quality laws and sparse monitoring networks hinders epidemiological studies on the health effects of air pollution, which are often not sustained over long periods, even in countries like South Africa that have air quality laws. South Africa's data quality for air pollution, such as PM levels, is poor, leading to an inability to report on PM2.5 levels as required by SDG 11.
The WHO has noted uncertainties and research gaps in the field, such as understanding the health impacts of mixtures of multipollutant exposures. Researchers have employed various statistical methods and, more recently, machine learning techniques, with a focus on unsupervised machine learning like k-means clustering, although such studies are scarce globally. There is a need for stronger integration and collaboration in the fields of epidemiology, biostatistics and data science.
Hence, we want to host a workshop to communicate studies that employ artificial intelligence (AI) methods such as machine learning and deep learning in air pollution exposure assessment and epidemiology studies. The workshop aims to facilitate discussions about the practical application of these AI methods.
The aim of the workshop is furthermore to explore future collaborative initiatives as well as to enhance policymakers' understanding of AI and its methodologies. As technology increasingly shapes policy landscapes, equipping decision-makers with nuanced insights becomes imperative.
Additionally, it is essential to address the inherent limitations of AI methods. Initiatives focusing on ethical considerations, potential biases, and areas where human judgment remains irreplaceable will contribute to responsible and ethical AI use. We believe joint discussions on how we can collectively contribute to these initiatives, will promote an extension of our collaboration beyond the confines of SASUF.
Anna Oudin and Janine Wichmann will be the hosts of the workshop.
1. Brief overview of data science and public health (15 min) (Sean Patrick, University of Pretoria, in person)
2. Examples from South Africa (30 min):
Application of unsupervised machine learning in an air pollution epidemiology study (Nandi Mwase, University of Pretoria PhD graduate, online)
Application of deep learning methods to predict air pollution (Rirhandzu Novela, University of Venda, online/in person)3. Example from Sweden (15 min):
The use of machine learning in environmental epidemiology: Empirical real world evidence from a human cohort study (Anna Oudin, Lund University, in person)Improving air pollution forecasts using machine learning algorithms (Magnuz Engardt, Stockholm University, in person)
4. Group discussion (45 min)
4. Summary and way forward (15 min) (Anna Oudin and Janine Wichmann)
Keywords: Air pollution epidemiology, AI, machine learning, deep learning, exposure assessment
Zoom link:
https://lu-se.zoom.us/j/62001648656?pwd=TVlHTTFmYWF1bko0YVBJWXIvbWczdz09
Password: 787237