Uncovering Respiratory Effects of Short-term PM2.5 Exposure through Global Nocturnal Cough Monitoring

Cecilia Mascolo, Xia Tong, E. Carlsson, Mikael Kågebäck, Marco Montagna, John S. JiResearch Square 2025 · DOI

Abstract

Air pollution remains one of the leading environmental determinants of health, yet the short-term effects of fine particulate matter (PM₂.₅) on early respiratory symptoms are poorly characterized owing to the limited availability of high-resolution, population-scale data. Enabled by global mobile sensing through a widely used application (Sleep Cycle) and AI algorithms that automatically detect coughs from nocturnal audio recordings, we derived nighttime cough frequency across 34 cities in 12 countries on five continents over 500 consecutive days, conducting the first multi-city assessment linking daily PM₂.₅ exposure to nocturnal respiratory symptoms. Single-city generalized additive models, multi-city panel regressions adjusting for meteorological variables and influenza rates, and quasi-experimental analyses consistently showed that same-day PM₂.₅ exposure was associated with higher cough frequency (relative risk = 1.004, 95% CI 1.001–1.008 per 10 μg/m³), with amplified effects during influenza peaks. The exposure–response curve was monotonic but nonlinear, with measurable risks evident even below the World Health Organization 24-hour guideline (15 μg/m³) and sharply increasing risks above 50 μg/m³. These findings provide real-world, population-level, multi-city evidence of acute respiratory effects from short-term PM₂.₅ exposure, reveal the absence of a safe threshold, and underscore the urgency of stringent air-quality interventions to protect respiratory health. This study also demonstrates how large-scale, AI-based mobile sensing can transform real-time environmental epidemiology by enabling population-level monitoring of early respiratory symptoms worldwide.

Introduction

While the long-term mortality risks of air pollution are well-documented, tracking the immediate effects of daily fluctuations in fine particulate matter (PM₂.₅) on early respiratory symptoms has historically been challenging. Traditional epidemiological studies often rely on severe clinical outcomes, such as hospital admissions or insurance claims, which fail to capture the broader "evidence gap" of sub-clinical respiratory irritation in the general population. This study addresses this gap by leveraging high-resolution, global mobile health (mHealth) data to monitor nocturnal coughing—a sensitive early indicator of respiratory distress.

Global study overview: linking daytime PM2.5 exposure to nighttime cough across 34 cities.

Methods

The research team integrated several large-scale datasets to analyze the relationship between air quality and respiratory health over 500 consecutive days across 34 cities in 12 countries.

AI-Driven Cough Detection

Using the Sleep Cycle smartphone application, the researchers employed a validated machine learning model—combining convolutional and recurrent neural layers—to automatically detect and timestamp cough events from nocturnal audio recordings. This provided a non-invasive, population-scale measure of respiratory symptoms.

Statistical Framework

  • City-Specific Modeling: Generalized Additive Models (GAMs) were used for each city to estimate the association between PM₂.₅ and cough frequency, adjusting for temperature, precipitation, and seasonal trends.
  • Global Synthesis: Fixed-effects panel regressions pooled data globally while controlling for time-invariant city-level factors like baseline health and socioeconomic status.
  • Quasi-Natural Experiment: To strengthen causal inference, the authors analyzed a massive wildfire event in Los Angeles in early 2025, which provided an exogenous surge in pollution levels.

Key Results and Insights

The study provides robust evidence that short-term PM₂.₅ exposure leads to an immediate increase in nocturnal coughing across diverse global environments.

The "No Safe Threshold" Discovery

A critical finding was the monotonic, nonlinear exposure-response curve. Respiratory risks were measurable even at PM₂.₅ concentrations below the World Health Organization’s (WHO) 24-hour guideline of 15 µg/m³, with risks increasing sharply above 50 µg/m³. This suggests that current safety guidelines may not fully protect against acute respiratory irritation.

Pollution-Infection Synergy

The impact of PM₂.₅ was found to be significantly amplified during peak influenza periods. This synergy suggests that viral infections and particulate matter work together to exacerbate airway inflammation, leading to higher rates of coughing than either factor would cause alone.

Synergistic effect: Influenza outbreaks amplify the respiratory impact of air pollution.

Immediate Impact and Global Consistency

The strongest associations were observed at "lag 0" and "lag 1," indicating that the respiratory system reacts to poor air quality almost instantly. The study found a Relative Risk (RR) of 1.004 (95% CI 1.001–1.008) per 10 µg/m³ increase in same-day PM₂.₅. Positive associations between pollution and coughing were consistent across 24 of the 34 cities studied, spanning five continents and various socioeconomic contexts.

Discussion and Future Directions

This study demonstrates the transformative potential of AI-based mHealth sensing in environmental epidemiology. By moving from reactive clinical data to proactive, real-time symptom monitoring, researchers can better understand the immediate health impacts of environmental stressors.

Limitations

The study notes several limitations, including potential selection bias (app users may be younger and healthier) and the use of outdoor air quality stations as a proxy for indoor exposure. Additionally, individual-level confounders like pre-existing conditions were not tracked.

Future Work

Future research could integrate individual-level data from wearable sensors and indoor air quality monitors to further refine these models. Such advancements will be crucial for developing personalized health alerts and more effective public health interventions in an increasingly polluted world.

Summary updated

Abstract

<title>Abstract</title> Air pollution remains one of the leading environmental determinants of health, yet the short-term effects of fine particulate matter (PM₂.₅) on early respiratory symptoms are poorly characterized owing to the limited availability of high-resolution, population-scale data. Enabled by global mobile sensing through a widely used application (Sleep Cycle) and AI algorithms that automatically detect coughs from nocturnal audio recordings, we derived nighttime cough frequency across 34 cities in 12 countries on five continents over 500 consecutive days, conducting the first multi-city assessment linking daily PM₂.₅ exposure to nocturnal respiratory symptoms. Single-city generalized additive models, multi-city panel regressions adjusting for meteorological variables and influenza rates, and quasi-experimental analyses consistently showed that same-day PM₂.₅ exposure was associated with higher cough frequency (relative risk = 1.004, 95% CI 1.001–1.008 per 10 μg/m³), with amplified effects during influenza peaks. The exposure–response curve was monotonic but nonlinear, with measurable risks evident even below the World Health Organization 24-hour guideline (15 μg/m³) and sharply increasing risks above 50 μg/m³. These findings provide real-world, population-level, multi-city evidence of acute respiratory effects from short-term PM₂.₅ exposure, reveal the absence of a safe threshold, and underscore the urgency of stringent air-quality interventions to protect respiratory health. This study also demonstrates how large-scale, AI-based mobile sensing can transform real-time environmental epidemiology by enabling population-level monitoring of early respiratory symptoms worldwide.
BibTeX
@inproceedings{mascolo2025uncovering,
  author = {Cecilia Mascolo and Xia Tong and E. Carlsson and Mikael Kågebäck and Marco Montagna and John S. Ji},
  title = {{Uncovering Respiratory Effects of Short-term PM2.5 Exposure through Global Nocturnal Cough Monitoring}},
  booktitle = {Research Square},
  year = {2025},
  doi = {10.21203/rs.3.rs-7977267/v1},
}
Figure