Influence pathways of noise exposure on people's negative emotions and health across different activity contexts: A neural network-based double machine learning approach

Abstract

Noise is a major global environmental issue that raises concerns about both mental and physical health. However, few studies have investigated the mediating role of emotions in the pathways linking noise exposure to health outcomes. Additionally, many studies have overlooked the varying effects of noise across different activity contexts. Most importantly, previous research has predominantly relied on correlational analysis, offering limited evidence of causality. In this study, we utilize real-time data from an environmental health survey of 800 residents in Hong Kong collected between 2021 and 2023 and apply a neural network-based Double Machine Learning model to estimate the pathways through which noise influences emotional states and health outcomes. Our findings reveal that (1) noise during travel significantly heightens real-time annoyance, while noise at home primarily increases real-time stress; (2) annoyance strongly contributes to headaches, whereas stress predominantly leads to insomnia and fatigue; and (3) noise at home directly triggers insomnia and fatigue, whereas noise during travel not only directly causes insomnia and headache but also indirectly exacerbates insomnia, fatigue, and headache through heightened annoyance. In contrast, noise in the workplace and outdoors has a limited impact on insomnia, fatigue, and headaches. This study provides valuable insights into the pathways through which noise influences negative emotional states and, subsequently, health outcomes, offering a methodological framework for unraveling the "black box" of environmental health relationships.