Daily Stress Recognition from Mobile Phone Data, Weather Conditions and Individual Traits (1410.5816v1)

Abstract: Research has proven that stress reduces quality of life and causes many diseases. For this reason, several researchers devised stress detection systems based on physiological parameters. However, these systems require that obtrusive sensors are continuously carried by the user. In our paper, we propose an alternative approach providing evidence that daily stress can be reliably recognized based on behavioral metrics, derived from the user's mobile phone activity and from additional indicators, such as the weather conditions (data pertaining to transitory properties of the environment) and the personality traits (data concerning permanent dispositions of individuals). Our multifactorial statistical model, which is person-independent, obtains the accuracy score of 72.28% for a 2-class daily stress recognition problem. The model is efficient to implement for most of multimedia applications due to highly reduced low-dimensional feature space (32d). Moreover, we identify and discuss the indicators which have strong predictive power.

• The paper presents a novel person-independent model for daily stress recognition combining mobile phone data, weather conditions, and personality traits.
• Using a Random Forest classifier, the multifaceted model achieved 72.28% accuracy in classifying stressed vs. non-stressed states, outperforming single-stream models.
• This non-intrusive approach enables scalable real-time stress monitoring applications leveraging smartphone data for mental health management and personalized wellness.

Daily Stress Recognition from Mobile Phone Data, Weather Conditions, and Individual Traits

The paper "Daily Stress Recognition from Mobile Phone Data, Weather Conditions, and Individual Traits" presents a data-driven approach to automatically recognizing daily stress by leveraging ubiquitous mobile data, environmental factors, and personality analysis. The authors propose a person-independent model that combines these diverse data streams and achieves a commendable classification accuracy.

Methodology and Key Findings

The research adopts a multi-faceted approach to stress recognition by merging three distinct data sources:

1. Mobile Phone Data: The system captures user behavior patterns from call logs, SMS interactions, and Bluetooth proximity data, representing dynamic indicators of social behavior and routine activities.
2. Weather Conditions: Ambient environmental factors, such as temperature, humidity, and wind speed, are analyzed for their potential impact on stress levels.
3. Personality Traits: The Big Five personality traits (Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism) serve as stable psychological indicators that modulate the effects of environmental and behavioral stressors.

The paper formulates stress recognition as a binary classification problem—differentiating between stressed and non-stressed states—achieving an accuracy of 72.28% using a Random Forest classifier. The classifier utilizes a reduced feature space of 32 dimensions, providing computational efficiency for real-world multimedia applications. The thorough feature extraction and selection process ensures robust generalization across the participant population, underscored by leave-one-out cross-validation.

Comparative Analysis

An insightful comparison is made between the multifactorial model and simpler models based on single or paired data streams. Results unequivocally demonstrate the superiority of the multifaceted approach, with single-stream models (utilizing only mobile, weather, or personality data) falling short in predictive accuracy. This substantiates the hypothesis that stress is a complex interplay of multiple transient and stable factors.

Implications and Future Directions

This paper importantly imparts a non-intrusive and scalable method for stress detection, leveraging the inherent capabilities of smartphones. It opens avenues for real-time stress monitoring applications which could significantly impact mental health management, workplace productivity, and personal wellness. These tools would empower users with timely insights to manage their stress proactively, paving the way for advancements in personalized health technologies and context-aware applications.

In the future, the methods could be expanded through the integration of additional data streams, such as social media activity or biometric data, for further advancements in generalizability and precision. Additionally, extending the methodology to longitudinal studies could enhance the understanding of chronic stress patterns and their societal implications.

The multifaceted model not only offers promising applications in health psychology and behavioral sciences but also underscores a broader vision of using mobile technologies as pervasive instruments for human-centric sensing and analysis.