myCityMeter / whatIbreathe

Abstract

Long-term exposure to PM2.5 and ambient noise is associated with mild cognitive impairment (MCI) in older adults. We present myCityMeter, a pollution exposure management tool for older adults and their caregivers, and whatIbreathe, which tracks cumulative PM2.5 exposure. A mixed-methods study (n=321) revealed that although 94% of participants were concerned about air pollution, less than 10% checked it weekly — motivating the need for seamless, personal monitoring tools.

Keywords: air pollution, PM2.5, older adults, cognitive impairment, IoT, mobile sensing, wearable, UbiComp

1. Motivation

Bridging the Gap Between Concern and Action

Epidemiological studies show that long-term exposure to PM2.5 and ambient noise is positively associated with mild cognitive impairment (MCI) in older adults — the transitional stage between normal aging and early dementia. Yet existing monitoring solutions operate at the neighborhood or county level (e.g., 4 EPA PM2.5 stations for all of Chicago), not at the personal exposure level.

94%concerned about adverse health effects of air pollution

<10%checked air pollution levels at least once a week

60%never checked air pollution levels around them

4EPA PM2.5 stations covering all of greater Chicago

Survey results showing pollution monitoring practices

Figure 3. Most respondents reported never checking air pollution levels. Among those who did, top sources were news (38%), mobile apps (22%), and government websites (21%). (HCII 2020)

2. myCityMeter System

A Hybrid Personal Exposure Monitoring System

myCityMeter combines a personal mobile sensing module with Chicago's Array of Things (AoT) city-wide sensor network to give users personal, geolocated PM2.5 and noise readings with high spatial resolution.

myCityMeter system overview showing hardware and app

Figure 1. System overview: mobile sensing module (Raspberry Pi 3b+ + Plantower PMS 5003), RESTful API middleware accessing AoT sensor nodes, and Samsung Note 8 companion app. (UbiComp 2018)

Figure 3. Phone app: (a) pollution lookup map, (b) exposure dashboard with PM2.5 and noise levels, (c) suggested actions based on current readings. (UbiComp 2018)

The system records six exposure metrics: average PM2.5 and noise over the last 24 hours and year, plus nighttime noise (10pm–6am). These metrics reflect the specific epidemiological findings on environmental risk factors for MCI. Users can also take the Self-administered Gerocognitive Examination (SAGE) through the app, with caregiver access to scores and location.

Technical architecture: Raspberry Pi 3b+ · Plantower PMS 5003 sensor · Bluetooth/Wi-Fi/LTE fallback chain · Amazon AWS cloud · RESTful API middleware · Voronoi-weighted interpolation across AoT nodes · Android (Java).

3. whatIbreathe

Tracking Cumulative PM2.5 Exposure

whatIbreathe extends myCityMeter with a focus on cumulative, longitudinal exposure tracking — helping users understand not just current air quality but how much polluted air they have breathed over time. This directly addresses the epidemiological concern: it is long-term, cumulative exposure that drives cognitive risk, not any single bad-air day.

Demo

The whatIbreathe demo video shows real-time PM2.5 monitoring integrated with GPS-geotagged historical exposure data, helping users connect their daily routes with their pollution burden.

4. User Survey (n=314)

Understanding Monitoring Practices & Preferences

An online survey (27 questions, April–November 2019, metro Chicago area) examined current air pollution monitoring practices and design preferences for wearable monitoring devices.

Survey results on device form-factor preferences

Figure 5. Willingness to monitor personal PM2.5 exposure: 84% wanted long-term exposure data, 82% willing to have a home device, 59% willing to wear one. Most valued portability (76%). (HCII 2020)

Figure 6. Top form-factor choices: bag accessory (74%) and wristwear (42%). Participants least preferred neck and shoe-worn devices. (HCII 2020)

5. Technology Probe Study (n=7)

Qualitative Probe with Vulnerable Users

Seven participants from vulnerable groups (older adults, people near industrial areas, those with cardiovascular or respiratory conditions) explored a working keychain technology probe — a $30 Plantower PMS7003 sensor on a Raspberry Pi 3b+ with a companion Android app — and discussed whether and how they would use it.

Technology probe device and participant exploring it

Figure 7. The technology probe: Plantower PMS7003 sensor mounted on Raspberry Pi 3b+, enclosed in a keychain form factor, with a companion smartphone application for real-time PM2.5 visualization. (HCII 2020)

6. Key Findings

Five Themes from Qualitative Analysis

Theme 1

Smaller, Lighter, Modular

Participants wanted a system usable both on-the-go and stationary at home — smaller when portable ("half of this"), mountable when at home, ideally wrist or bag form factor.

Theme 2

To Monitor or Not to Monitor

Attitudes split between wanting to take action vs. anxiety about knowing: "To me it's almost better not to know" — and concern about constant alerts inducing panic.

Theme 3

Access to Longitudinal Data

Users wanted long-term exposure data to correlate with sick days, compare neighborhoods, and understand personal health triggers — not just current readings.

Theme 4

Ambivalence toward Lifestyle Changes

Limited actionable options outside the home created ambivalence. Indoors: filtration systems. Outdoors: little beyond masks. The sociotechnical problem of short-term vs. long-term reward remains unresolved.

Core Design Tension

The limited options for controlling personal exposure appeared to demotivate use of monitoring tools. Future systems must offer tangible, personalized steps — e.g., predicting how changing a running route or time of day reduces PM2.5 exposure by a specific amount.

**Table 3.** User requirements and associated design tradeoffs for personal air pollution monitoring systems (HCII 2020, Table 3).
User Requirement	Design Tradeoff
Small, light, carriable	Accuracy may be unreliable at certain concentrations; smaller battery means frequent charging; skin emissions may interfere
Good accuracy	Calibration models require additional sensors, making the device bulkier; seasonal recalibration needed
Longitudinal exposure data	Data may generate anxiety/helplessness without means to act; long-term exposure–health links still vary by demographics
Tangible actions	GPS always-on raises privacy concerns; location logging creates security vulnerability