Predicting social anxiety disorder based on communication logs and social network data from a massively multiplayer online game: Using a graph neural network

• DOI: 10.1111/pcn.13804
• Publication Date: 2025-03-11T13:12:23Z
• Authors: Kenji Yokotani; Masanori Takano; Nobuhito Abe; Takahiro A. Kato
• Keywords: Social network (sociolinguistics)

Aim Social anxiety disorder (SAD) is a mental that requires early detection and treatment. However, some individuals with SAD avoid face‐to‐face evaluations, which leads to delayed detection. We aim predict based on their communication logs social network data from massively multiplayer online game (MMOG). Method The study included 819 users of Pigg Party, popular MMOG in Japan. Participants completed the Japanese version Liebowitz Anxiety Scale (LSAS‐J) withdrawal scale (hikikomori) questionnaire. scoring ≥60 LSAS‐J were classified as having SAD, while those <60 not (non‐SAD). A total 142,147 users' 613,618 edges Party used input whether participants had or non‐SAD. Graph sample aggregated embeddings (Graph SAGE) was utilized graph neural model. Results Individuals more likely be socially withdrawn physical community (hikikomori), fewer friends, spent less time other virtual houses, showed lower entropy visitation times MMOG. Based data, SAGE model predicted an F1 score 0.717. Conclusion include indicators interpersonal avoidance behaviors, typical SAD; this suggests potential use digital biomarkers for SAD.