Participative observatory for the surveillance of individual exposure to air pollution relating to health – POLLUSCOPE

The implementation of the target mobile crowdsensing system will tackle many challenges related to metrology, data collection protocol, environmental health, data management and analysis. Polluscope brings together specialists in environmental, health, geosciences and computer science, which covers all the aspects of this multi-disciplinary project. The system will be deployed and tested on the ground. The measurement campaign is to equip volunteers (diseased - suffering from asthma or COPD - and healthy subjects) with lightweight sensors including a GPS receiver, an accelerometer, sensors of diverse pollutants, temperature and humidity. The collected data will be exploited for validating of two scenarios: the one is an epidemiological study, while the second corresponds to the citizen science schema. A remote server will provide advanced capabilities of analysis of these data. The key challenge of this server is to efficiently process and analyse the sensor data flow. For this purpose, data science technology is crucial. This includes efficient filtering, geographical contextualisation, advanced statistical analysis and data mining. This will enable microenvironment detection, user’s activity recognition, comparing profiles of pollution exposures, and so on.

Polluscope is expected to result in significant added-value contribution to the state-of-the-art and revolutionary social impact, due to the interdisciplinary yet complementary skills of the consortium members. More specifically, the expected research outcome and project novelty can be summarised as follows: 1. To develop a novel community-based participatory sensing system specifically designed for the observation of individual exposure to environmental risks. 2. To provide powerful and efficient techniques for integrating, matching, and enriching individual’s exposure data. 3. To cover both indoor and outdoor air pollution (dwelling, work, transportation, etc.) while automatizing context detection. 4. To extract useful, practical, and rich knowledge from collected data by applying data mining and pattern recognition techniques. 5. To provide techniques and tools to enable the identification of pollution side effects on individual’s health and in particular the respiratory and cardiac functions. 6. To allow for an online access information system for the public to get up-to-date pollution information as well as publishing their own measurement.

The benefits of Polluscope will extend beyond its running period by the establishment of a long-term running, and evolutionary framework, giving rise to a new concept, namely, a community-based participatory observatory for pollution and exposure. This is expected to bridge the gap between the emergent technology of sensors and the decision support for both the concerned organizations and the participants themselves. This perspective will drastically change the way individual’s exposure to air pollution and exposure variability are measured, perceived, and evaluated. The project purpose is a proof of concept. However, a large-scale deployment would probably lead to further research and development work.

B. Languille, V. Gros, N. Bonnaire, C. Honoré, C. Debert, L. Gauvin, S. Srairi, A. Gorin, B. Chaix, I. Annesi-Maesano, M. Chachoua, C. Ray, K. Zeitouni. Observatoire participatif de l’exposition individuelle à la pollution de l’air : première phase de tests et sélection des capteurs, communication orale, Journée scientifique du SIRTA, juin 2017.
B. Languille, V. Gros, N. Bonnaire, C. Honoré, C. Debert, L. Gauvin, S. Srairi, A. Gorin, K. Zeitouni, 2018, « First stage of the Polluscope project: selection and assessment of portable air quality sensors », Oral presentation, Air Quality Conference, Barcelone, March 2018.
R. Haque, Y. Taher, and K. Zeitouni. Towards AirCare: A Highly Scalable Data Management System for Mobile Sensor Driven Exposure Analysis to Air Pollution. IEEE ComSoc Newsletter. 2017.
A. Mustapha, Y. Taher, and K. Zeitouni. Towards Rich Sensor Data Representation: Functional Data Analysis Framework for Opportunistic Mobile Monitoring. 4th International Conference on Geographical Information Systems Theory, Applications and Management, GISTAM 2017.

The quality of the air we breathe is a central concern of individuals living in urban and suburban areas. Millions of people are exposed every day to air pollution at high levels. The impact of such pollution on the human health is extremely alarming. Particularly, WHO and IARC have classified air pollution, including fine particles, as certain carcinogenic. Understanding the totality of exposures to air pollutants over the course of our daily life is a key concern to reduce the risk of some major diseases. However, the ability to acquire high-quality, relevant, and useful individual’s exposure data is challenging. Currently available air pollution fixed station networks allow to only account for background air pollution and less frequently proximity air pollution from road traffic. As a result, the measurements made through this kind of network typically provide the average exposure to air pollution in a specific geographical zone. In particular, they fall short to quantify the real individual’s exposure with respect to his/her indoor/outdoor daily life activities in different settings, such as transport, work, dwellings, etc.
Nowadays, an increasing number of wearable and lightweight environmental sensors have emerged, enabling a continuum measurement of the real personal exposure anywhere at anytime. Such an evolution has been the main enabler of providing new solutions for data acquisition, namely community-based participatory sensing where citizens contribute data to the system with the purpose of sharing events of interest within the community. This technology has recently gained a great interest among the actors of environmental science in public, associative, and private sectors, while stimulating a wide range of research projects worldwide.
Building on top of such a technology evolution, Polluscope aims at bringing together experts from environmental, metrology, epidemiological, and data sciences while providing methodologies, techniques, and tools – expected to drastically change the way individual’s exposure and exposure variability are measured, perceived, and evaluated. Such measurements will not only consider gaseous pollutants (Ozone, NO2), but also particulates (via particulate matter and black carbon) and those typical of indoor environments (VOC) – providing a representative overview of the air pollution. Gaining such enriched insights into individual’s exposure will contribute towards reducing individual risks of some diseases by changing their behavior. This will end up in a solid, invaluable, and vital societal impact namely, saving life and improving the individual well-being.
To achieve these objectives, a novel infrastructure for real individual’s exposure data acquisition, processing, and analysis will be develope. For this to be done, several scientific and technical challenges come into the picture. The data are collected at a high frequency and might be massive and noisy. Therefore, the system must be able to process them efficiently, while taking into account both their velocity and their uncertainty. More importantly, it has to offer microenvironment and user’s activity recognition, through integration with external spatiotemporal resources. An efficient data collection and analysis will provide an insightful knowledge on individual’s exposure over his/her daily life activities, and will enable conducting analytical queries, novel risk assessment modeling, mining and comparing profiles of pollution exposures, and so on. Therefore, it is evident that a robust, efficient, and powerful data science technology is crucial.
Lastly, Polluscope will be evaluated under real-world use cases. Several type of population will be targeted by the data acquisition campaign. Both diseased and healthy subjects will be involved to conduct an epidemiological study relating air pollution exposure to health on the one hand, and volunteer participants for the crowd sensing on the other hand.