Using Mobile Sensing to Test Clinical Models of Depression, Social Anxiety, State Affect, and Social Isolation Among College Students

Philip I. Chow, Karl Fua, Yu Huang, Wesley Paul Bonelli, Haoyi Xiong, Laura E. Barnes, Bethany A. TeachmanJournal of Medical Internet Research 2017 · DOI

TL;DR

This study leverages mobile sensing (GPS and in-situ affect ratings) to explore the complex interplay between emotional states, social anxiety, depression, and social isolation in college students. The findings highlight the feasibility of using this technology to gain a more nuanced understanding of how daily emotional experiences relate to real-world behavior like staying home, offering a promising avenue for mental health research and intervention.

What Problem Does This Paper Address?

Traditional research on mental health, particularly concerning depression and social anxiety, often relies on static, self-reported data. This approach has limitations in capturing the dynamic, moment-to-moment nature of emotional experiences and their connection to actual behavior in everyday life. While we know that negative affect and social withdrawal are linked to these conditions, understanding the precise temporal relationships and how they manifest in real-time has been challenging. The paper aims to bridge this gap by integrating passively collected GPS data (as a proxy for social isolation via "homestay") with actively collected in-situ state affect ratings from mobile phones. This allows for a more ecologically valid examination of how transient emotional states, moderated by trait-level depression and social anxiety, influence an individual's tendency to stay home.

Methods

The researchers employed a novel approach combining mobile sensing with self-report measures.

  • Mobile Sensing:
    • GPS Tracking: A custom Android app (Sensus) was installed on participants' smartphones to continuously collect GPS location data. This data was used to determine "homestay" – the percentage of time spent at home, serving as a proxy for social isolation.
  • In-Situ State Affect Ratings: The Sensus app also prompted participants up to six times daily at random intervals to rate their current positive and negative affect on a visual analog scale. This captured transient emotional states in real-time.
  • Self-Report Measures:
    • Social Anxiety: Assessed using the Social Interaction Anxiety Scale (SIAS) at a baseline laboratory session.
  • Depression: Assessed using the 7-item depression subscale of the Depression, Anxiety and Stress Scale (DASS-21) at a baseline laboratory session.
  • Data Analysis:
    • The study explored four different statistical models to examine the temporal relationships between affect and homestay:
  • Associations between changes in affect within short (up to 4-hour) windows and homestay during that same window.
  • Associations between mean-level affect within a day and homestay on the same day.
  • Associations between mean-level affect within a day and homestay on the following day.
  • Associations between mean-level affect within a day and homestay on the previous day.
  • Generalized mixed-effects models were used to account for individual differences and daily variations. Both continuous (log-transformed ratio of time spent at home) and dichotomous (likelihood of being at home) measures of homestay were analyzed. Social anxiety and depression were entered as moderators in these models.

Datasets

  • Participant Data: 72 undergraduate students from a southeast US university participated.
  • Data Collection:
    • Self-report measures of social anxiety and depression were collected at baseline laboratory sessions.
  • In-situ state affect ratings and GPS location data were collected passively and actively via a custom mobile app (Sensus) for up to two weeks.

No new benchmarks were introduced; the study utilized existing self-report scales and standard GPS data processing techniques.

Key Insights and Ideas

  • Feasibility of Mobile Sensing for Mental Health: The study successfully demonstrates that integrating passively collected GPS data with actively collected in-situ affect ratings via mobile phones is a feasible and valuable approach for studying mental health in everyday life.
  • "Homestay" as a Proxy for Social Isolation: Time spent at home, derived from GPS data, serves as a practical and ecologically valid indicator of social isolation, a key behavior associated with depression and anxiety.
  • Temporal Dynamics Matter: The research highlights the importance of examining temporal relationships. Different models (e.g., short-term affect changes vs. daily averages) reveal distinct patterns of association between affect and homestay.
  • Social Anxiety as a Key Moderator: Social anxiety consistently emerged as a significant moderator, influencing the relationship between affect and homestay, particularly in predicting future behavior.
  • The "Double Whammy" Effect: For individuals high in social anxiety, experiencing higher negative affect or lower positive affect on a given day was associated with a greater likelihood of staying home the following day. This suggests a potential feedback loop where negative emotional states can lead to increased social withdrawal.

Key Results

  • Main Effects:
    • Higher social anxiety was consistently associated with more time spent at home, both in 4-hour windows and during the workday (10 AM - 6 PM).
  • Higher negative affect was associated with more time spent at home on the same day.
  • Higher positive affect was associated with a lower likelihood of being at home on the same day.
  • Higher positive affect was associated with more time spent at home the next day (contrary to initial hypotheses).
  • Higher depression was associated with a lower likelihood of being at home during the workday (10 AM - 6 PM) (unexpected finding).
  • Interactions:
    • A significant interaction was found: among individuals with higher social anxiety, greater negative affect on a given day predicted a higher likelihood of staying home the following day.
  • Similarly, among individuals with greater social anxiety, lower mean-level positive affect on a given day was associated with a greater likelihood of staying home the next day.
  • Statistical Significance: The authors note that the number of significant findings (8 out of 48 possible effects) was highly unlikely to be due to random chance (p < 0.001), suggesting the observed patterns are meaningful.

Limitations and Future Work

  • Correlational Nature: The study's findings are correlational, meaning causal relationships cannot be definitively established. Future work could involve experimental manipulation of affect or homestay to explore causality.
  • Sample Characteristics: The study was conducted on undergraduate students, who may have more uniform daily routines than a general population. Replication with a community sample and clinical populations is recommended.
  • Self-Report Limitations: While in-situ ratings were used, they are still self-reports. The authors acknowledge that depression and negative affect are not perfectly aligned measures.
  • Other Factors: Potential influences like alcohol use, sex, and race were not fully explored due to statistical power limitations.
  • Technical Issues: Software bugs and compatibility issues led to some data loss, with an average of only 2.5 prompts per day being received.
  • Future Directions:
    • Investigate the "double whammy" effect further with more frequent in-situ sampling of emotions and thoughts.
  • Explore passive sensing of social interaction (e.g., via Bluetooth) and physiological correlates of affect (e.g., heart rate, skin conductance).
  • Develop more nuanced measures of social isolation beyond just "homestay."
    • Incorporate contextual data (e.g., weather, social environment) to build more complex predictive models.
  • Examine the impact of sex and race on these relationships.

Summary updated

Abstract

Results demonstrate the feasibility and utility of modeling the relationship between affect and homestay using fine-grained GPS data. Although these findings must be replicated in a larger study and with clinical samples, they suggest that integrating repeated state affect assessments in situ with continuous GPS data can increase understanding of how actual homestay is related to affect in everyday life and to symptoms of anxiety and depression.
BibTeX
@article{chow2017using,
  author = {Philip I. Chow and Karl Fua and Yu Huang and Wesley Paul Bonelli and Haoyi Xiong and Laura E. Barnes and Bethany A. Teachman},
  title = {{Using Mobile Sensing to Test Clinical Models of Depression, Social Anxiety, State Affect, and Social Isolation Among College Students}},
  journal = {Journal of Medical Internet Research},
  year = {2017},
  doi = {10.2196/jmir.6820},
}
Figure