Mental health and wellbeing are one of the most challenging issues in modern society. For example, moderate stress can help a person in many beneficial ways to confront a challenge. On the other hand, excessive stress, a common phenomenon in our society, can cause overall negative health and wellbeing impact, such as increasing susceptibility to infection and illness, affecting a diverse range of physical, psychological and behavioral conditions (i.e., anxiety, depression, and sleep disorders, or decreasing job productivity). Furthermore, mental disorders such as depression and schizophrenia, if not monitored and treated timely, can lead to further degradation of the person’s mental health and wellbeing. The ability to measure stress levels or mental health could enable better self-management of one’s behavioral choices in ways that might be intervened timely.
While various methods have been proposed for automatic stress or mental health detection using wearable or mobile phone data, it is far from solved. Besides, it is an understudied research question of reproducibility, due to the lack of a proper publicly accessible dataset and baselines. This workshop introduces to the research community the publicly accessible datasets with both anonymized hand-crafted features and deep features for in the wild stress and mental health sensing challenges, including:
The goal of this workshop: is to inspire ideas and collaborations, raise awareness of reproducibility problems in modeling wearable data in the wild, and drive the research frontier. Publicly sharing the datasets, including both features and baselines, will accelerate research activity such as multimodal wearable data processing and modeling, handling missing data, and personalization. We will accept research papers about detection of stress and mental health using wearable sensors and technical solutions using our open stress datasets.