Breathing Easy with AI

Air pollution has long been a global health concern due to its detrimental impact especially on children's well-being. The World Health Organization estimates that air pollution causes more than 2.2 million deaths in the Asia Pacific region every year, and scientific evidence solidly connects air pollution to adverse birth outcomes, infant mortality, lung damage, asthma, cancer, neurological disorders, and childhood obesity. Unfortunately, impoverished communities often face higher exposure to air pollution, putting them at a heightened risk of health issues like pneumonia, which remains a leading cause of child mortality. 

Given the escalating threats posed by climate change and the ever-growing levels of air pollution that children are exposed to, this moment offers an incredible opportunity for harnessing the potential of big data and AI analytics to guide prevention and response efforts that combat air pollution, strengthen health systems worldwide, and safeguard children's health. However, when it comes to responding to air quality issues, one of the major blocks is the lack of access to real-time data.  

This is where Thinking Machines comes in. With seed and acceleration investments from the UNICEF Venture Fund alongside close collaboration with the Frontier Data Tech Node in UNICEF’s East Asia and Pacific Regional Office, the company developed the AI model Air Quality as a part of its larger Artificial Intelligence for Development (AI4D) initiative. Alongside Geowrangler and Relative Wealth Mapping, Air Quality is a part of a catalog of AI models that streamline data science for practitioners at the intersection of machine learning and development.  

With its focus on public health, Air Quality provides real-time information at a sub-district level in Thailand, leveraging data from low-cost and reference-grade sensors alongside satellite-derived air quality indices. Early testing indicates Air Quality's potential for transformative impact, offering governments and organizations a cost-effective, accessible solution with real-time insights, eliminating the need for expensive technology. 

“Air quality monitoring is critical in informing health and environmental programs, especially with millions of children suffering from toxic air pollution, but data from air quality ground sensors are often sparse.

 

In Thailand, Thinking Machines and the UNICEF East Asia and Pacific Regional Office (EAPRO) created an ML model to estimate haze (particulate matter 2.5) even in areas with limited air quality monitoring equipment.” 

 

Pia Faustino, Director for Social Impact and Sustainability 

The true transformative power of AI4D lies in its Open Source nature, where the data generated is available for all public health authorities, development partners and caregivers alike, often at minimal or no cost. Additionally, the democratization of data facilitates faster responses to air quality issues, offering replicability in various contexts. 

“Thinking Machines built interpretable machine learning (ML) models that are low-cost, globally replicable, and have the deployment potential for development sector use cases through the Artificial Intelligence for Development (AI4D) Initiative. 

 

By publishing our code and technical resources on the AI4D ML web catalogue, we hope that other data teams can replicate and build on our research.” 

Pia Faustino, Director for Social Impact and Sustainability 

Alongside partners including Takeda Pharmaceutical Limited Company, UNICEF Venture Fund invests in start-ups in low- and middle-income countries developing innovative tech solutions that strengthen localized health systems to meet global health challenges.

In the fight against air pollution, access to real-time air quality information is crucial in protecting the health of children and communities. The collaboration between UNICEF and Thinking Machines is a significant step towards leveraging the potential of AI and data science to improve health outcomes for children everywhere.