JMIR Publications

ABSTRACT

Background:

Medical students face academic and psychological pressures. Online forums have become vital platforms for peer support and information exchange, offering an authentic window into their concerns and community dynamics. However, few studies have analyzed these interactions longitudinally based on real-world data from online social platforms.

Objective:

This study aims to integrates topic modeling and social network analysis to investigate the evolving concerns and influence structures within the bulletin board system of medical university over a seven-year period.

Methods:

We collected all posts and comments from the “Medical Center” board of Peking University BBS forum between 2017 and 2023. Latent Dirichlet Allocation (LDA) was used for topic modeling to identify primary discussion topics. Social network analysis (SNA) was also employed to construct interaction networks. And degree centrality, eigenvector centrality, and betweenness centrality were calculated to identify key users and analyze network evolution over time.

Results:

The corpus included 4,639 posts and 58,942 comments. The annual post numbers fluctuated over years, reaching its lowest point in 2019, and peaking in 2022. Seven major topics of concern were identified, with feedback on living facilities (22.1%) being the most prevalent, followed by information inquiry (21.6%) and medical education (19%). The SNA revealed a generally sparse network structure (median degree centrality: 0.000745, IQR: 0.000521-0.001815; median eigenvector centrality: 0.004527, IQR: 0.001946-0.009820; median betweenness centrality: 0.000011, IQR: 0.000000-0.000102), within which a small set of influential nodes with high degree, eigenvector, or betweenness centrality played distinct and critical roles. Subnetwork analyses revealed year-to-year turnover of core nodes alongside a small subset of persistent influencers.

Conclusions:

This research demonstrates that online forums are organic components of the medical education ecosystem, reflecting students' primary needs for improved living conditions, efficient information services, and robust peer support. The integration of topic modeling and SNA provides a powerful, data-driven framework for educational administrators to enhance resource allocation, optimize communication strategies, and foster targeted support by collaborating with influential student leaders.