Automatic Stress Detection in Working Environments from Smartphones' Accelerometer Data: A First Step (1510.04221v1)

Abstract: Increase in workload across many organisations and consequent increase in occupational stress is negatively affecting the health of the workforce. Measuring stress and other human psychological dynamics is difficult due to subjective nature of self- reporting and variability between and within individuals. With the advent of smartphones it is now possible to monitor diverse aspects of human behaviour, including objectively measured behaviour related to psychological state and consequently stress. We have used data from the smartphone's built-in accelerometer to detect behaviour that correlates with subjects stress levels. Accelerometer sensor was chosen because it raises fewer privacy concerns (in comparison to location, video or audio recording, for example) and because its low power consumption makes it suitable to be embedded in smaller wearable devices, such as fitness trackers. 30 subjects from two different organizations were provided with smartphones. The study lasted for 8 weeks and was conducted in real working environments, with no constraints whatsoever placed upon smartphone usage. The subjects reported their perceived stress levels three times during their working hours. Using combination of statistical models to classify self reported stress levels, we achieved a maximum overall accuracy of 71% for user-specific models and an accuracy of 60% for the use of similar-users models, relying solely on data from a single accelerometer.

• The paper presents an innovative method for detecting occupational stress using smartphone accelerometer data, achieving 71% accuracy with user-specific models.
• The methodology employs non-invasive sensor monitoring combined with statistical models like Naive Bayes and decision trees to classify stress levels.
• The research offers practical insights for workplace health, advocating for the integration of stress monitoring into everyday technology to mitigate burnout.

Analyzing Automatic Stress Detection Using Smartphone Accelerometer Data: An Expert Overview

This paper investigates a novel approach to monitoring occupational stress through smartphone accelerometer data. Recognizing the increased prevalence of work-related stress, it proposes an objective methodology to assess stress levels using non-invasive technology, thus providing an innovative avenue in health informatics research and practice.

Methodology Overview

The authors conducted an eight-week paper involving 30 healthy participants from two different organizations. Participants were equipped with smartphones to capture accelerometer data continuously throughout their workdays, reporting perceived stress levels thrice daily using a self-assessment questionnaire. The focus on accelerometer data is distinctive due to its minimal invasiveness, low power usage, and reduced privacy concerns compared to other sensors like cameras or microphones.

The paper aimed to validate if statistical models can effectively classify stress based on accelerometer data alone. Features were extracted from raw accelerometer data, and various statistical models were employed, including Naive Bayes and decision trees, supplemented by a feature selection technique. A noteworthy facet is the exploitation of user-specific, general, and similar-user models to accommodate the diversity of individual stress responses and data sparsity.

Numerical Results and Model Evaluation

The paper achieved a notable accuracy rate of 71% for user-specific models, demonstrating the potential of personalized predictive analytics in stress detection. This performance is slightly reduced to 60% with similar-user models, which indicates a promising foundation for predicting stress in scenarios where individual-specific data is limited. General models yielded lesser accuracy, which reflects the complex and individualized nature of stress responses that a one-size-fits-all model cannot robustly capture.

Further evaluation using ordinal classification, which maps stress levels as an ordered set ({low, medium, high}), did not enhance prediction accuracy, suggesting limitations due to the few distinct class categories.

Practical and Theoretical Implications

The practical implications of this research are significant, especially in the field of occupational health and well-being. By integrating stress detection into commonly used devices like smartphones or fitness trackers, organizations could proactively monitor employee stress, potentially reducing burnout and associated health issues.

Theoretically, the research elucidates the feasibility of using minimal sensor data for complex psychological state detection, advocating for further exploration into multimodal sensing and contextual awareness to augment current methodologies. This research, steeped in the real-world application, bridges the gap between technology capability and human behavioral insight.

Future Directions

The paper posits several avenues for future exploration, including extending the duration and scale of studies to enhance data robustness and predictive model accuracy. The integration of additional sensory inputs, such as textual data during phone interactions, could refine stress analysis granularity. Further, the authors suggest expanding model adaptability to evolving personal behavior patterns over time, thus advancing toward dynamic and personalized stress management systems.

In summary, this paper offers a methodical approach to stress monitoring, emphasizing objective, scalable, and ethically conscious methodologies, paving the way for innovations in automated health tracking in professional settings.