Avatars' Social Rhythms in Online Games Indicate Their Players' Depression

Area of Science:

• Digital biomarkers and mental health research.
• Human-computer interaction in gaming.
• Computational psychiatry and behavioral analysis.

Background:

• Gameplay data, or digital biomarkers, offer insights into player characteristics.
• An avatar's online behavior is hypothesized to correlate with its player's mental health status.

Purpose of the Study:

• To estimate the depressive state of online game players using their avatar's behavioral logs.
• To investigate the relationship between gameplay patterns and depression.

Main Methods:

• Utilized behavioral logs and depression questionnaire data from 3,361 Pigg Party players in April and May 2020.
• Trained a time series model on April data to predict depression in May data.
• Analyzed activity patterns within a 24-hour cycle.

Main Results:

• Individuals without depression showed more consistent 24-hour activity cycles compared to those with depression.
• A time series model trained on initial data predicted depression with 0.3–0.4 accuracy in subsequent data.
• Model performance indicated challenges in predicting depression from sparse time series data.

Conclusions:

• Avatar behavior in online games can be a potential indicator of player mental health.
• Current predictive models struggle with sparse behavioral data, necessitating further development and refinement.
• Future research should focus on improving models for accurate mental health state estimation from digital biomarkers.