OpenAQ Use Case: UNICEF Venture Fund-backed Startup Building Global Air Pollution Model to Map Children’s Exposure to Air Pollution
Children are particularly susceptible to air pollution’s insidious effects. They breathe more rapidly than adults, taking in more polluted air. Due to having smaller bodies, they are proportionally exposed at higher levels. Their developing organs are more vulnerable to pollution. And due to their height, they are face level with pollutants that settle lower in the air and pollutants, such as vehicle exhaust, emitted close to the ground. Where biomass fuel is used for cooking, children are usually carried on their mother’s back, directly exposing them to indoor air pollution. In a 2018 report, Air Pollution and Child Health, WHO estimated outdoor and indoor air pollution resulted in 543,000 deaths in children under the age of 5 years in 2016.
Recognizing that they could play a part in addressing this health crisis, data scientists Christina Last and Prithviraj Pramanik recently formed a startup, AQAI, to build an open-source machine learning tool to predict pollution exposure for 1.8 billion children globally. Based in Kolkata, India, AQAI (Legal name: Formative Resilience Know-How Pvt Limited) is one of nine startups (out of 450 applicants) awarded investment funding from the UNICEF Venture Fund in April 2022.
Christina is a postgraduate student at MIT. Previously, she was a Research Data Scientist at the Alan Turing Institute (the UK’s Artificial Intelligence Institute) in London, England. Prithviraj is a PhD student in Computer Science in the National Institute of Technology in Durgapur, India. Prithviraj and Christina are also both Fulbright fellows.
“Prithviraj and I met in 2020 via the UNICEF team as volunteers working on air quality data,” says Christina, “and the research we did led to the idea to create the company.” Prithviraj and Christina recognized the power to leverage artificial intelligence (AI) and machine learning (ML) to further social good. AQAI aims to fuse together real-time satellite imagery and ground-based sensing to replace stale air quality data being used in health impact assessments. The tool will evaluate exposure to particulate matter (PM) in places where there is a low density of AQ sensors. Currently, AQAI is working alongside UNICEF Belize and exploring opportunities with UNICEF Mongolia in modeling pollution exposure for children and schools. “We are building on the work completed by UNICEF Monitoring and Evaluation Specialists to engage children in the issue of air pollution, helping them understand local trends in pollution exposure using our machine learning models,” shares Christina.
Their pilot project is in a region in Latin America and the Caribbean, where they have tested AQAI’s modular software stack and programmable API, which incorporates data ingestion, model training, prediction, and processing local data — including that from OpenAQ’s data platform — to create local models. According to Christina, “OpenAQ’s data aggregation platform has made our work possible. We ingest historic and real-time information from OpenAQ globally, allowing us to predict local pollution concentrations to inform our work with UNICEF country offices.” Prithviraj also points out the harmonization and ease of data acquisition from OpenAQ via the API, which gave them the ability to focus more on downstream tasks rather than the tedium of looking for these data themselves. “Structured air quality data is hard to acquire from global agencies that follow diverse API structures for querying. When we initially started the exploratory data analysis, OpenAQ simplified the data acquisition from federal regulatory monitors.”
Prithviraj and Christina will present their project at the 2022 American Geophysical Union Fall Meeting, one of the world’s largest gatherings for Earth and space scientists.
Read more about the UNICEF Venture Fund and its 2022 cohort here: UNICEF invests in AI-powered education and health system solutions | UNICEF Office of Innovation.
Listen to The Turing Podcast: Data Science for Social Good: Predicting air pollution in a post-COVID world? featuring Christina Last, Prithviraj Pramanik and Dr. Subhabrata Majumdar: https://turing.podbean.com/e/solve-for-good-is-the-air-harming-children-s-health/
Twitter handles: @last_christina, @PrithvirajP01, @AirQualityAI