Human exposure to air pollution is associated with higher levels of heart rate variability, electrodermal activity and a worsened mental wellbeing, according to researchers at Nottingham Trent University.
In this study, researchers from the Smart Sensing Lab at Nottingham Trent University have developed and observed (for the first time) the relational impact between the urban environment, personal characteristics, behavior, and mental wellbeing.
The approach pulls in data from different modalities and shows in real-time, instant impact. We utilized the ecological momentary assessment (EMA) method by sampling participants current behaviors and physiology in real-time within the current environment.
Participants partook in an experiment following a pre-specified route around Nottingham Trent University Clifton Campus. Each participant was given an environmental monitoring kit, wearable (to monitor physiological changes) and smartphone app (to self-report wellbeing), as depicted in Figure 1. All data collected by the three devices was time-stamped and geo-tagged.
The researchers found that there was a direct correlation between Electrodermal Activity (EDA) and Heart Rate Variability (HRV) being noticeably impacted by the level of Particulate Matter (PM) in the environment. Previous research has shown that EDA and HRV are linked to the sympathetic nervous system and provide a good indication of stress and wellbeing indicating worsening pollution has a negative impact on wellbeing.
The team then employed mathematical and statistical approaches for the exploratory analysis stage which included: variable correlations, PCA factor maps, variable importance, and Pearsons R correlation coefficient to measure the association between the two variables.
In addition, to summarize the dynamic sensing patterns, we utilized heat maps and spatial visualizations on the collected data. In particular, through heat maps shown in Figure 2, we can observe the changes in moving from the university campus towards a busy polluted main road (upper right of the map).
The increase of red within the map indicates a very high reading from the environmental and physiological sensing equipment. In particular, each participant once at this location experienced a heightened level of HRV, EDA, Noise and PM2.5. Overall, when near a busy polluted road HRV and EDA increased, suggesting that pollution induces stress.
To further understand the relationship between the environment and wellbeing, we utilized machine learning and deep learning models (including Deep Belief Networks) in order to classify the wellbeing states collected in the experiment. Using this approach, we achieved 80.83% accuracy classifying five self-reported states of well-being using the environment and physiological data.
The team also found through analysis on the environmental and physiological data that when participants labelled their wellbeing as very negative, there were extremely high levels of PM1, PM2.5 and PM10 in the environment.
Figure 3 shows PM2.5 along with each user self-reported state of wellbeing demonstrating worsening wellbeing with increased pollution
Thomas Johnson, PhD Candidate in Pervasive Computing and Data Science at Nottingham Trent University, said: ‘For the first time we have shown a real-time, instant approach which demonstrates the effect that exposure to urban environmental pollutants can have on physiology and wellbeing. We believe this is the first step towards understanding the link between how changes within the environment can impact our wellbeing.’
The research team highlight that this first of its kind study could have interesting implications on air quality in the future and how places can be designed to improve mental wellbeing. The combination of Data Science and Statistical Analysis paves the way for effective monitoring and understanding of the environment and the impact this has on improving overall life quality.
