MY_AIR

Project Synopsis

Air pollution has been identified by the World Health Organization (WHO) as the world’s largest single environmental health risk [1]. In the U.S., 166 million people live in areas with unhealthy air [3]. The U.S. Environmental Protection Agency’s AirNow program has been providing air quality information and associated health risk indicator - Air Quality Index (AQI) - to the public since 1998 [6]. The data source is the EPA’s air monitoring station network. However, more than 42 million people reside in populated places farther than 40 km from the nearest PM2.5 monitor [7]. In addition, the published AQI or pollutant concentration values do not reflect individuals’ health risk to pollution exposure. This is because personal exposure risk also depends on the person’s microenvironment (indoor, outdoor, in-vehicle), activity (sitting, sleeping, running, walking, biking etc.), and physiology (age, gender, health condition).

To this end, we are building the MY-AIR (Monitor Your Air-pollution Risk) software package for PM2.5 in this project. MY-AIR consists of a smartphone app and enabling models and algorithms on the server end. Different from existing mobile apps on air quality, MY_AIR applies advanced deep learning techniques to interpolate current and predict near-future air pollutant levels at fine spatial and temporal scales, and provides personalized AQI (PAQI) by combining an individual’s activity and location information.

Figure 1 is the architecture of MY-AIR. There are three key components:

- MY-AIR user interface. When an individual uses the app for the first time, s/he is asked to provide his/her age, gender, physical health conditions, notification frequency, and agreement to be tracked by MY-AIR (by allowing MY-AIR to get access to the GPS readings). After becoming a registered user, s/he receives hourly PAQI values throughout the day via MY-AIR and alerts for unusual events (e.g., unusually high level of a hazardous pollutant in the near future at specified location). Users are also given options to rate the app and provide feedback to improve the app.

- MY-AIR algorithms. Running in the background of the app are recordings of smartphone sensors data, the Semantic Location and Activity Recognition (SLAR) algorithms, and the PAQI estimation algorithm.

- Server. The pollutant modeling produces hourly local concentrations, which will then be used to compute

A prototype of MY_AIR will be tested with sickle cell patients from the UI Health Sickle Cell Center.

Live demo coming soon... Stay tuned.

Figure 1. Architecture of MY-AIR Figure 2. MY_AIR mobile app interface

Investigators:

Student Research Assistants:

  • Graduate students: Hui Shen, Marco Miglionico, Giovanni Monna, Marco Mele, Luca Dibattista

  • Undergraduate students: Simrah Shaik, Adam Tayabali, Tarush Gupta, Syed Raza, Jasper Gabriel

  • High school student: Alec You (Whitney Young)

Funding Source:

This project is funded by the University of Illinois Discovery Partnership Institute (DPI) seed grant. It is led by Dr. Lin as PI, Drs. Robert Molokie of College of Medicine and Ouri Wolfson of Computer Science as co-PIs. The project goes from June 2019 to June 2020.

IRB protocol:

  • # 2018-0375 “Enviromental Exposure and Sickle Cell Disease”

Relevant Publications:

[1]. Miglionico, M., 2018. A deep learning framework for air quality monitoring, MS thesis, Department of Computer Science, University of Illinois at Chicago.

[2]. Stenneth, L., Wolfson, O., Yu, P., Xu, B., 2011. Transportation Mode Detection using Mobile Devices and GIS Information, Proc. of the 19th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM GIS),Chicago, IL, pp. 54-63.

[3]. Ma, S., Wolfson, O., Xu, B.,2014. UPDetector: Sensing Parking/Unparking Activities Using Smartphones, Proc. of the 7th ACM SIGSPATIAL International Workshop on Computational Transportation Science, pp. 1-10.

[4]. Liu, J., Wolfson, O., Yin, H., 2006. Extracting Semantic Location from Outdoor Positioning Systems, International Workshop on Managing Context Information and Semantics in Mobile Environments (MCISME), pp. 1-8.

[5]. Monna, G.C., 2018. MY-AIR Project: Study on Semantic Location and Activity Recognition Algorithms for iOS Systems, MS thesis, Department of Computer Science, University of Illinois at Chicago.

[6]. Vallamsundar, S., Lin, J., Konduri, K., Zhou, X., Pendyala, R. (2016) A Comprehensive Modeling Framework for Transportation-induced Health Exposure Assessment, Transportation Research Part D: Transportation and the Environment, 46, 94-113.

[7]. Vallamsundar, S., Lin, J., Chang, Y.T. (2016) Estimating Externality of Population Health Exposure to Emissions From Transportation Sources, International Journal of Shipping and Transportation Logistics, Vol. 8, No.6, 632-652.

[8] Lin, J., Wolfson, O.+ (2020) MY-AIR: A Personalized Air-pollution Information Service, invited paper, IEEE International Conference on Smart Data Services, October 19-23, 2020.