It is a rare person that cannot relate to the concept of stress, we all might define it differently given our personal experience, it might be stressful to stay upright on a packed subway train on your work commute, it might be driving on the freeway, or being with in-laws during the Holidays. Whatever your definition, at a physiological level our sympathetic nervous system is hard at work helping us navigate our fight or flight response.
From an evolutionary perspective, this response was useful, see a lion who thinks you are dinner, run away to preserve your life- you get the picture. Back in the 1950s, Hans Selye’s defined stress as “the nonspecific response of the body to any demand.” With the advent of technology, the plethora of sensors that assess movement, heart rate, sweat (skin conductance) is allowing us to evolve our assessment of stress to enable us to develop digital phenotypes well beyond questionnaires like the gold standard Perceived Stress Scale PSS). Many of these markers are reasonable measures of stress in a laboratory setting and the value of sensors in day to day life is the continuous measurement in real life scenarios.
A new paper in Nature Digital Medicine by Elena Smets and colleagues from Electrical Engineering-ESAT, Leuven, in Belgium examine the use of wearable data to create digital phenotypes for daily stress. Up to this publication, the authors point out that studies using sensor data were based on small samples, usually around fifty participants. The current research named SWEET, (Stress in the Work Environment) has over one thousand healthy participants, assessed over a five-day period which included baseline data collection. Participants were encouraged to go about their daily lives; regular assessment of physiological measures was taken in addition to psychological assessments including perceived stress, depression and anxiety.
Findings show distinct phenotypes that warrant further study. The authors divided the groups in low, medium and high-performance ranges which allowed them to look at differences between participants over the five days and to calculate the dynamic range of their physiological features, noted as low, (small dynamic range) or high (large dynamic range).
Those in the high-performance group had a more significant variation in their stress measures (86% no stress, 12% light stress, and 2% high stress) as compared to low performance (6% no stress, 45% light stress, and 29% high stress) this may have impacted the results as there were also differences in the reporting frequencies of the self-reported psychological assessment (26 times vs. 31 times in the 5-day period).
Participants varied on their psychological baselines and demographics, the group with a more blunted physiological stress-reactivity (small dynamic range) reported a less healthy lifestyle and higher depression, anxiety, and stress scores than the more responsive group. These findings suggest that self-reported poor health and high depression scores are negatively associated with physiological reactivity.
This paper is a welcome addition to our understanding of sensor and questionnaire derived data. The different patterns emerging to those with a high or low dynamic range (wide/narrow physiological response) speaks to the need for larger datasets to look at individual data over time to personalize recommendations to support health interventions. It will be interesting to see how the apple watch’s new ECG feature will fare in our every changing daily milieu of stress. Certainly how we perceive our stress levels and psychological states impacts our physiology in positive and negative ways.
Thanks for reading – Trina
(Opinions are my own)
Nature Digital Medicine