ML/AI, Data Science, and Exposure Science: Lessons Learned from Engagement with Communities and Public Health/Policy Planning for the Future

BACKGROUND: 

A grand challenge facing environmental public health practitioners (and related fields) is addressing disparate pollution impacts and create more just infrastructure futures. Simultaneously, artificial intelligence (AI), machine learning (ML), and data science (DS) are disrupting public health and our ways of defining and addressing these problems. While researchers can contribute to understanding problems and ideating solutions by leveraging these methods in novel ways, engagement with communities is essential to develop complete representations of the problems and designing and implementing corrective actions. 

OBJECTIVES: (1) Discuss the work being done in understanding the sources of disparate exposure to air pollution throughout Ohio. (2) Describe the application of the discussants technical work and how this engages impacted communities to produce better, more relevant insights. (3) Share their successes and failures of these approaches. (4) Synthesize these findings and approaches to understand how ML/AI and DS can be leveraged to inform public health and policy interventions to create environmental quality – and health outcomes – for all.

DESCRIPTION: Dr. Jeffrey Bielicki will discuss ongoing communityengaged work in the Columbus Metropolitan Area to shape the transition to electrified mobility and home energy use in historically underserved communities. Dr. Daniel Gingerich will discuss efforts that integrate data science with life-cycle thinking to inform technology selection and decision-making for energy and water infrastructure. Dr. Mike Rayo will discuss the development of the MaRC/PERCC toolkit, ML/AI-based tools to collect and analyze community feedback for the infrastructure design and planning process. Dr. Darryl Hood will discuss aspects of the above via an exemplar using CPRG (Climate Pollution Reduction Grant). Dr. Hood will also discuss potential guardrails which are necessary in the use of ML/AI algorithms in healthcare systems to reduce bias and improve population health and policy. Dr. Huyen Le will discuss collaborations with local agencies (Mid-Ohio Regional Planning Commission and Franklin County Public Health) to deploy low-cost air quality sensors and predict highly resolved spatio-temporal PM2.5 estimates for exposure assessment research and equitable planning practice. Finally, Dr. Gingerich will facilitate a conversation among Discussants to highlight common themes how their studies can inform policy design and implementation.