Towards Continuous Monitoring of Well-being

In recent years, the detrimental effects of prolonged stress and burnout have gained significant attention, particularly in work settings. Burnout, characterized by excessive and prolonged stress, takes a toll on individuals' physical and mental well-being. Recognizing the impact of long-term stress on overall wellness, researchers propose the integration of Intelligent Environments (IE) to improve well-being by monitoring and detecting stress at an early stage. IE refers to surroundings where intelligent agents control and adapt various environmental factors, enhancing individuals' experiences. One effective approach involves the use of wearable sensors that continuously monitor physiology and behavior without causing discomfort, enabling users to interact naturally with their environment while gaining insights into their health. Despite advancements in wearable technology, ensuring accurate measurements of physiological and mental states remains challenging. Therefore, we present a study that compares real-time self-reported arousal annotations with Electrodermal Activity (EDA), aiming to validate EDA as a real-time stress indicator. Additionally, the research explores the potential of measuring stress on a continuous scale rather than discrete states, which could contribute to the development of more accurate and sensitive measures of well-being and facilitate better management of stress-related health issues over time.

Despite advancements in wearable technology, ensuring accurate measurements of physiological and mental states remains challenging.

To evaluate the relation between Electrodermal Activity and self-reported arousal we made us of the Continuously Annotated Signals of Emotion (CASE) dataset. This dataset allowed for the research since it contains various physiological recording, including EDA, as well as a continuous annotation of self-reported valence and arousal. We extracted ‘non-specific’ Skin Conductance Responses (NSSCRs) from the EDA signal and employed a sliding window to calculate their frequency and amplitude. Simultaneously, we processed self-reported arousal data. We used the Pearson coefficient to evaluate the relationship between self-reported arousal and NS-SCRs.

The samples are categorized into three groups dependent on stimuli: low, mid/high, and high arousal. A one-sampled t-test showed the statistical significance of the observed associations between self-reported arousal and NS-SCRs for the mid/high, and high arousal conditions. These findings suggest a strong association between the measured NS-SCRs and the self-reported arousal levels, specifically in situations where arousal levels were moderately high or high. This indicates the potential for NS-SCRs to serve as a reliable indicator of self-reported arousal in these specific contexts.

Findings suggest a strong association between the measured NS-SCRs and the self-reported arousal levels

From a societal perspective, allowing IE to monitor stress has the potential to revolutionize our understanding and management of stress-related issues. By providing real-time data on an individual's stress levels, it enables personalized interventions and preventive measures, promoting overall well-being and mental health. Nevertheless, several limitations of our study need to be considered. Currently, results are based on data collected in a lab. Validating the accuracy and reliability of EDA as a continuous measure in real-world conditions is crucial to ensure the effectiveness of interventions. Hence, future research must determine the generalisability of our results and ultimately show the utility of EDA monitoring in IE.

‍