Using wearable biosensors and ecological momentary assessments for the detection of prolonged stress in real life

Abstract

Emerging efforts toward prevention of stress-related mental disorders have created a need for unobtrusive real-life monitoring of stress-related symptoms. We used ecological momentary assessments (EMA) combined with wearable biosensors to investigate whether these can be used to detect periods of prolonged stress. During stressful high-stake exam (versus control) weeks, participants reported increased negative affect and decreased positive affect. Intriguingly, physiological arousal was decreased on average during the exam week. Time-resolved analyses revealed peaks in physiological arousal associated with both self-reported stress and self-reported positive affect, while the overall decrease in physiological arousal was mediated by lower positive affect during the stress period. We then used machine learning to show that a combination of EMA and physiology yields optimal classification of week types. Our findings highlight the potential of wearable biosensors in stress-related mental-health monitoring, but critically show that psychological context is essential for interpreting physiological arousal detected using these devices.

Changes in large-scale neural networks under stress are linked to affective reactivity to stress in real life

“…This measure is novel in capturing inter-individual differences in positive affect changes in response to in-the-moment stress. We focus on positive affect as our previous work has shown that stress seems to have a bigger impact on positive rather than negative affect 31 . This may be due to respondents showing greater variability when answering positive affect items, which is reflected in the skewed distribution of negative affect.…”

“…Within the scope of the current paper, only subjective stress and positive affect measures are used from the EMA data. Full details and results of the EMA weeks are reported in previous work, and the full questionnaire set can be found in the associated GitHub directory 31 .…”

“…We used a measure of stress combining three types of stress: Event-related, activity-related, and social stress. Positive affect was measured using a 4 item questionnaire adapted from previous EMA studies ( See 31 for full results and details). Participants showed a significant increase in the numbers of beeps during the exam week in which stress was above their overall average (Odds Ratio=-1.67, SE=0.14, t-stat=6.033, p<0.001, Figure 2A) in addition to lower overall positive affect (β=-0.77, SE=0.11, t-stat=-6.80, p<0.001, Figure 2B) compared to the control week.…”

Getting stress-related disorders under control: The unique opportunity of neurofeedback-based therapy

“…For other stress-related disorders, pinpointing the optimal time for an early intervention might be more difficult, as they are not necessarily triggered by a single event, but can develop more gradually. However, research has suggested that the transition into a mental disorder is foreshadowed by a gradual deterioration of resilience with a slowing-down of stress recovery during repeated exposure to stressors (Nelson et al, 2017), and recent efforts into the detection of early warning signs using ecological momentary assessments and wearables were already successful in predicting dailylife stress (Tutunji et al, 2022) as well as the risk of symptom transitions (Wichers et al, 2020). Targeting individuals before such critical transition points with prospective neurofeedback training will strengthen their resilience to future stressors and has the potential to prevent a progression into stress-related mental illness (see also Figure 3b).…”

Automated mood disorder symptoms monitoring from multivariate time-series sensory data: Getting the full picture beyond a single number

“…stress) and recruited general populations (e.g. college students) were assessed with self-reported questionnaires and commercial wearable devices [6, 69, 78]. Only few previous endeavors specifically addressed mood disorders using clinician assessments.…”