Imec’s SWEET study collects world’s largest dataset on stress detection

Work-related stress is common in modern-day society; however, while everyone experiences occasional stress, chronic stress can have a significant and long-term impact on emotional and physical well-being, and it can cause depression, anxiety disorders and professional burnout. According to the American Institute of Stress, stress is estimated to cost the US economy approximately 300 billion dollars a year (due to absenteeism and productivity loss). In Europe, the annual cost of stress is estimated at 514 billion euros in productivity loss and 63 billion euros in direct healthcare costs.

Currently, the most widespread method to detect stress is by the means of questionnaires. However, these questionnaires are subjective, time-consuming, and are conducted on a spot-check basis only. To lower the risk of reaching the threshold of chronic stress, it is important to identify stress signals immediately and to provide personalized, just-in-time feedback so that the individual can employ correction strategies to decrease his/her stress level. Imec’s wearable technology in combination with advanced algorithms to analyze the data collected can play an important role in this.

Imec’s large-scale research, the SWEET study, tracked more than 1,000 participants. First, participants’ baseline stress levels were determined via validated psychological questionnaires. They were then provided a wrist band and a wireless ECG patch, which they wore continuously for five days. The ECG patch monitored their heart rate and heart rate variability along with acceleration (movement). The wrist band, fortified with advanced algorithms, measured skin conductance, skin temperature and acceleration (movement). Physiological stress symptoms were then supplemented by contextual data collected through the participants’ smartphones, such as GPS data, phone activity and noise level, and self-reported information.

“Our SWEET study is unique as it is the first large-scale study to use multiple wearables to establish the link between physiological stress symptoms and self-reported stress in real-life. Utilizing wearables in this research generated complex and sophisticated data sets, taking real-life contextual factors into account that helped us better understand periods of stress and its indicators,” explained Elena Smets, imec.ichange researcher and PhD student at KULeuven. The first results, for instance, already indicated that participants’ average heart rate variability correlated with their perceived stress levels.

Participants were queried 12 times a day via a smartphone app to evaluate their self-reported stress levels and to answer multiple-choice questions about their daily activities, food and drink intake, sleep quality, and digestive processes. In addition, they also completed the Montreal Imaging Stress Task, a 20 minute-stress test that allowed the researchers to calibrate participants’ stress levels with their personal physiological symptoms.