Breathing Space

Poor air quality (AQ) costs the planet $8.1 trillion annually and is the leading factor in the global burden of diseases, affecting 93% of children’s health. Long-term exposure to fine particulate matter is estimated to cause ∼8 million excess deaths annually. The effect is felt by the poorest and the youngest, with 1 in every 4 deaths under five years related to environmental risks. In India, the air pollution crisis is particularly acute,  as ambient pollution concentrations result in the premature demise of around 10.5 million individuals annually. Of the top 50 polluted cities in the world, India has 39 such cities, emphasizing the need for evidence-based research in which participants in pollution events can be observed in real-time. 

But in a Low-Middle Income Country (LMIC), measuring air pollution remains challenging due to costly infrastructure and trained personnel. Changing industrial and transportation emissions sources and natural sources such as wildfires and weather patterns results in wide geographic variations and ambient pollution concentrations, increasing the logistical challenge and measurement cost. The problem is even more critical if we wish to understand individual exposure in such a setting. Staying in high-pollutant neighborhoods for a long time has desensitized the people, and they often are engaged in air pollution-inducing events themselves and/or take no remedial steps when they are faced with high levels of air pollutants. As youth (Ages: 15 - 24), being the agents of change in society, are at further risk given their prolonged exposure as air pollutant events have dramatically increased in the last 20 years compared to air pollution events in the last 40 years, especially in India.

Thus, given an area of medium-high air pollutant events in an LMIC like India, can the youth identify polluting events in the neighborhood and take proactive steps? If not, can we plan an intervention that can help in identifying these kinds of events such that they can take proactive steps?

Our solution is an innovative approach to measure pollution, identify sources and induce community-led efforts to understand pollutants and pollution in urban and semi-urban communities of India.

We do this in a three-pronged approach:

1. We will develop a localized AI-based approach to measuring pollution concentrations from open satellite imagery and a network of air pollution sensors to understand the baseline levels of community air pollution and track change throughout the intervention period of 8-10 months.

2. We will develop a “crowdsourcing game” for the youth population (15-24 years) in urban  and semi-urban areas of Telangana, India to log pollution sources within the community and tag when “community interventions are made.”  The game-based approach aims to incentivize the youth to educate their families on the harms of air pollution, the sources, and potential ways in which their approach towards air pollution can improve in and around their places of work, education and living.

3. We will run multiple dissemination workshops and refresher training in the communities to introduce air pollution as a health issue, safely identify sources of air pollution, and explain what to do if they see a community member contributing to air pollution. These workshops will be facilitated by our community partner The George Institute for Global Health India  (https://www.georgeinstitute.org.in/projects/community-action-for-health-in-india-cahi-from-the-national-rural-health-mission-to), who have wide-ranging experience engaging in community-based action in India.

1. Local AI-generated pollution predictions: We propose a hybrid approach where we take in the fine-grained accuracy from the ground-based sensors and pre-train models to fine-tune satellite data at greater accuracy. The models are at a 1 km x 1 km scale for regional levels. The major goal of this model would be to associate the data with the health effects and identify at-risk youths to target our intervention effectively. 

2. Crowdsourcing Game: Using the team’s prior experience developing tools to raise awareness of possible air pollution harm in the City of Philadelphia (https://www.breathingspacephilly.com/) we will develop a “crowdsourcing game” where participants can log air pollution sources and educate their community members on the dangers of air pollution. 

• Participants will be asked to tag various areas of a map to identify possible sources of air pollution such as open incineration, cook stoves, industrial facilities, and congestion. The participants will be asked to take a photo of the pollution source and describe the justification for logging the source, referring to the sights, sound and smell. After submitting the air pollution information, participants are able to see their submission along with all their other classmates, incentivizing their learning through interactive knowledge sharing. 

• A crowdranked algorithm then based on map locations that have been tagged by multiple players will indicate a high concentration of crowdsourced pollution sources, or multiple submissions of the same pollution source. The verified pollution sources are then incorporated into AQAI’s air pollution prediction algorithm, providing granular, high quality ground truth information on air pollution sources in regions of India which have sparse data.

For the first time, in the Urban & Semi-Urban areas of Telangana, there will be the opportunity to crowdsource pollution sources, initiate the process of behavioral change using a crowdsourcing game that incentivizes community engagement, and validate the community levels of air pollution using AI-based monitoring.

Why It Is Significant

Such a study is critical at this point in an LMIC there are there are disproportionately affected by air pollution as because of rapid urbanization, increased industrialization, and reliance on fossil fuels. Moreover, LMICs often lack resources to address air pollution. From our field visits, there is a requirement of multilingual support of the mobile application for an immersive experience, and it was seen in certain pockets there is limited internet connectivity. 

• Youth staying in an LMIC are more susceptible to the health impacts of air pollution: Children and adolescents are more vulnerable to the health effects of air pollution than adults, due to their developing bodies and respiratory systems. This makes educating and involving youth in efforts to reduce air pollution important.

• Youth are often more receptive to new ideas and behaviors: As they are at a stage of life where they are developing their own values and beliefs, they are often more open to new ideas and behaviors. This can make it an opportune time to introduce concepts like environmental stewardship and sustainable behavior.

• Youth can have a significant impact on their families and communities: Young people can serve as powerful advocates for environmental issues and inspire their families and communities to take action. This can lead to positive changes in behavior at a broader level.

• Youth are the future leaders: Engaging youth in environmental issues can help develop the next generation of leaders committed to creating a sustainable future. By investing in youth-focused interventions, we can help to ensure that the fight against air pollution continues long into the future.

Given our community partners, George Institute of Global Health, who are health practitioners and have a strong presence in the state of Telangana & the state contains the 2nd most polluted city in Southern India (Hyderabad), it gives us the unique opportunity to work close quarters with the community health experts. Not all in our target demography (Urban, Semi-Urban of Telengan, India) are well versed in English, hence to capture the variability in language in a diverse country like India, which has 31 individual languages which have more than 1 million native speaker as per the latest census (2011), we will implement the mobile application in English (10.7% of the Population), Hindi (57.1 % of the Population) & Telegu (7.8% of the Population) which can address more than half of the 1.4 Billion population of India. Moreover, the regions in question will require the support of native languages like Telegu (Urban & Semi-Urban of Telengana).

We are well-positioned to deliver on this mission as we have demonstrable experience deploying healthcare interventions in the communities most affected. We have already worked with UNICEF  Belize to deploy low-cost pollution sensors across Belize City. Alongside the municipal school network, we deployed, collected and collated data from “Air Quality Eggs” deployed by UNICEF. We visualized the pollution exposure patterns to engage community groups and schools.

Our team lead, Prithviraj Pramanik is the CEO of AQAI and is a Ph.D. Candidate and a Fulbright Fellow who has cost-effective urban air quality measurement techniques extensively. Mr. Pramanik has piloted technology deployment in West Bengal (Sundarban, Murshidabad, Paschim Burdwan), India, consistently working in non-Internet settings and required the development of mobile apps which could address that along with having multilingual support (https://itsforkit.github.io/). He has a proven track record of solution delivery working with UNICEF Latin America and UNICEF Regional office in Belize to deploy low-cost pollution sensors across Belize City. Alongside the municipal school network, he collated data from “Air Quality Eggs” deployed by UNICEF to engage community groups and schools (https://www.unicefinnovationfund.org/portfolio/aqai-formative-resilience).

Dr. Sreya Majumdar works as a Research Fellow at the George Institute for Global Health India, where she conducts qualitative and implementation research. She lives in Hyderabad, and has a long standing relationship with the pilot communities, as she works in states like Telangana, Kochi, Tamil Naidu in India and has conducted in depth interviews, and focused group discussions. Sreya has a Ph.D. in sociology and social anthropology, and she conducted ethnographic study in understanding birth professionals in India, with the goal of understanding their development as 'new professionals' in India. Sreya was a Fulbright fellow at the University of Texas at Austin, where she worked with communities to gain an understanding of their responsibilities in birthing practices.

Dr D Praveen (MBBS MD PhD), Program Director, Better Care India is a Senior Lecturer, Faculty of Medicine, UNSW Sydney.  He is a public health specialist with thorough knowledge of epidemiological study designs and having a keen interest in systems-based innovations to address inequities related to chronic diseases. His research experience is related to planning and managing large scale public health research projects and surveys. He is the head of the primary healthcare research at The George Institute India, based in Hyderabad. He has been awarded the competitive Australian Leadership Awards Scholarship in 2012 to pursue his PhD in the University of Sydney. His current focus is on health systems and in understanding the system-level barriers to address health system delivery.

There have been multi-level efforts from the Indian government to reduce air pollution from different fronts. The government in its flagship project National Clean Air Programme aims to reduce the PM2.5 & PM10 by 20% to 30% across the country comprehensively by 2024 with 2017 as the base year. Unfortunately as of September 2022, according to a public interest research and advocacy organization, the PM2.5 trends have hardly any difference between cities where the action was taken under this program and cities which were outside the purview of the program.

Your Connection to this Problem 

Figure 1: PM2.5 exposure across 3 District in Belize from the 32  Participants

1. UNICEF Belize: We are working with UNICEF Latin America and UNICEF Regional office in Belize to deploy low-cost pollution sensors across Belize City. Alongside the municipal school network, we are collating data from “Air Quality Eggs” deployed by UNICEF. We visualized the pollution exposure patterns to engage community groups and schools (https://www.unicefinnovationfund.org/portfolio/aqai-formative-resilience).

2. Philadelphia: We developed https://www.breathingspacephilly.com/  to raise awareness of possible air pollution harm in the City of Philadelphia. The community members can view crowdsourced air quality information across the city in real-time and also, in real-time, report it via this dashboard. For more details:(see full demo: https://docs.google.com/presentation/d/e/2PACX-1vQoS9LG9-KE_7J91-mq-ksptZHdrM-DmnBw-9NLUOTbTTcPlN7tr-HKvLaB5gAIvzv9wKd0qUmS4Ofg/pub?start=false&loop=false&delayms=3000) 

3. Sundarban: Our team members have prior expertise in field testing of post-disaster communication platforms in no/low-connectivity environments in Sundarban, the largest delta in the world and the rural areas of West Bengal, India https://itsforkit.github.io/. This work required the development of the App in Multilingual setting and work in areas where there is no Internet.

Market research

We have conducted thorough market research to increase the chances of success within our deployment in Telengana.

Project Logistics

Our solution will be deployed in urban and semi–urban areas in Telangana, India with support from community partners, the George Institute for Global Health. We will identify and recruit participants in these areas using the following criteria:

• Individuals between the age of 15-24 years

• High use of mobile phones within the community

• Historically high concentrations of air pollution

• Logistical ease of access to the community.

We will hold immersion and dissemination workshops for 2 days for participants to experience hands-on training sessions in using mobile applications to:

Understand why air pollution is harmful and what to notice (sights and smells) that could help identify air pollution sources.

After identifying pollution sources, participants will learn how to log air pollution incidents and describe the pollution source (collecting the date and location is completed automatically). 

How to effectively intervene, question certain behavior, and suggest pollution reduction strategies (e.g., opening a window when cooking indoors, cooking on smokey cookstoves outside rather than indoors).

We will develop a scoring system within the crowdsourcing game to measure the diversity of the measurements, e.g.: 

• The number of pollution sources captured

• Whether the participants was able to identify the source correctly

Whether the participants was able to take action safely, raise awareness among the individuals of the pollution source and its potential for harm. Youth will collect data as they move around and receive feedback on their progress through daily scores. To solicit feedback on the youth’s engagement, we will hold group discussions to encourage participation and reflectance. We want to undertake this effort for a minimum of 3 months to see how this goes and how their behaviors change with time. As we undertake this effort to validate our models, we will place low-cost air quality sensors in and around schools to validate our model. These sensors will be in place to capture the variability of the measurements. 

Community Partnerships

We are working with the George Institute of Global Health (https://www.georgeinstitute.org.in/)  as our community partner. The George Institute is a leading independent medical research institute established in India in 2007. Dr. D Praveen from the institute is our advisor, and their research Fellow Dr. Sreya Majumdar is one of the project investigators.

Ethical and Sustainability Considerations

The ethical considerations we are addressing are the following:

• Impact of location-enabled devices on youth - we will automatically disable location and only when conducting workshops will participants enable location.

• Impact of incentive schemes on outreach - we will conduct a thorough assessment of the ethical implications of paying members of the local communities.

3 Communities - 250 individuals

Given our community partners, George Institute of Global Health, who are health practitioners and have a strong presence in the state of Telangana & the state contains the 2nd most polluted city in Southern India (Hyderabad), the Solve Challenge us the unique opportunity to work close quarters with the community health experts. Not all in our target demography (Urban, Semi-Urban of Telengan, India) are well versed in English, hence to capture the variability in language in a diverse country like India, which has 31 individual languages which have more than 1 million native speaker as per the latest census (2011), we will implement the mobile application in English (10.7% of the Population), Hindi (57.1 % of the Population) & Telegu (7.8% of the Population) which can address more than half of the 1.4 Billion population of India. Moreover, the regions in question will require the support of native languages like Telegu (Urban & Semi-Urban of Telengana).

We built AQAI with a mission to help governments and humanitarian organizations reduce the impact of air pollution on the prevalence of acute respiratory diseases in children. Globally, 93% of children live in places where air pollution levels exceed World Health Organization guidelines. Now, with the full backing of UNICEF’s Office for Innovation, AQAI are the first venture to expose the locations where children are exposed to high pollution concentrations to justify investing in air quality improvement and carbon reduction projects.

To do this, AQAI have developed a production-grade, end-to-end machine learning pipeline to predict hyperlocal air pollution concentrations at any location on the planet, using a combination of satellite imagery global air pollution sensor networks. We are using this MVP to onboard our first five customers, consisting of UNICEF regional offices, National Governments and academic insitutions to gather data on potential cost and carbon savings from air pollution management.

For the next year our current goals are

1. Increase the accuracy of air quality predictions by incorporating more data sources and improving the AI algorithms in the target regions.
2. Connect with the Telengana government with the results and to provide real-time air quality data to residents and enable them to make more informed decisions about their health and outdoor activities.
3. Develop new features to the platform, such as personalized alerts and recommendations based on individual health factors and preferences.
4. Collaborate with other environmental organizations to advocate for better policies and regulations to improve air quality and expand to other regions.
5. Furthermore, in the next year we are deploying our AI-based technology in  Egypt, Mongolia and Belize, to help them monitor progress toward improving respiratory health, COP27 and Vision 2030 sustainability goals. 

Next five years, we aim to:

1. Achieve near-perfect accuracy in air quality predictions, thanks to ongoing improvements in AI algorithms and the availability of more and better data sources.

2. Expand to other other geographic areas, and have a sizable population using our platform to become a go-to resource for individuals, businesses, and governments seeking reliable air quality data, insights, and recommendations.

3. Partner with healthcare providers to leverage air quality data to improve patient outcomes and reduce healthcare costs.

4. Use the platform and data to inform research on the long-term health impacts of air pollution and advocate for systemic changes to reduce air pollution levels.

To measure progress towards Sustainable Development Goal impact goals for air quality, the following steps can be taken:

1. Establish Baseline Data: The first step is to establish baseline data for the air quality metrics, such as air quality index (AQI), particulate matter (PM),  and nitrogen oxides (NOx)from relevant agencies or measured using air quality monitoring equipment.

2. Set Targets: The next step is to set targets for air quality improvement based on SDG impact goals. For example, the target could be to reduce PM2.5 levels by 50% by 2030.

3. Monitor Progress: Progress towards the target region needs to be monitored regularly using our technology.

4. Analyze Data: The data collected from the users and our prediction engine can be analyzed regularly to determine if the targets are being met or if any corrective action is needed.

5. Implement Strategies: Strategies need to be implemented to improve air quality if the data indicates that the targets are not being met. Strategies could include promoting alternate waste management, cleaner fuel usage, implementing pollution control measures which are sustainable.

Our theory of change are the following:

1. To identify the major pollutants in the areas where the project will be conducted.

2. To be able to understand how people identify pollutants.

3. Through the orientation programs and workshops, to share the knowledge and make people aware of the major pollutants and the harm it is doing to the health and environment. 

AQAI is delivering the world's first open-source model measuring air pollution exposure anywhere in the world. We are developing a machine learning solution that helps predict where the highest concentrations of fine particulate matter are (PM2.5 or fine particulate matter is the air pollutant that poses the greatest risk to health globally) to augment accurate but limited air pollution data collected using ground sensors. A user can access air pollution data for 169,101,933 PM2.5 measurements and use our trained machine learning to predict air pollution concentrations for any 1km in 158 countries.

https://dl.acm.org/doi/abs/10.1145/3580279, Transaction on Sensor Network, 2023

https://arxiv.org/pdf/2103.12505, SIG CHI LBW 2021

The ethical considerations we are addressing are the following:

• Impact of location-enabled devices on youth - we will automatically disable location and only when conducting workshops will participants enable location.

• Impact of incentive schemes on outreach - we will conduct a thorough assessment of the ethical implications of paying members of the local communities.

Key Customers: Health Organization, Government

Key Beneficiaries: Health Organization, Government, Citizen

Key Product: Gamified App that will account for measuring the change in the users in understanding air quality.

SaaS, PaaS, Enterprise Level customization.

UNICEF Venture Fund- 100,000

Subak - 10000, 

Sandbox - 5000,  

NASA - 5000

Legatum - 3000

Software Sustainability Fellowship - 3000

OpenAQ - 1000, 

MIT IDEAS - 1000

Total = 1,28,000