With the rise of online health communities (OHCs), medical live streaming has emerged as a critical channel for physician–patient interaction. While prior research on live streaming has focused heavily on transactional outcomes, non-transactional engagement behaviors remain underexplored, especially in high-trust contexts like OHCs. This study aims to investigate how physicians’ emotional expressions influence patients’ different non-transactional engagement behaviors and how this relationship is moderated by disease type and professional title.
Drawing on emotions as social information (EASI) theory and signaling theory, this study conceptualizes physicians’ emotional expressions, specifically emotional valence and emotional variability as dynamic signals. Disease type and professional title are treated as static signals. Using data from a leading Chinese OHC, the study applies emotion recognition models to analyze 455 sessions about physicians’ emotional expressions and employs 2-way fixed effects regression to examine their impact on patient engagement.
The results reveal that physicians’ emotional valence significantly increases patients’ non-transactional engagement behaviors, while emotional variability exhibits a different effect on high-effort and low-effort engagement. Moreover, disease type strengthens the effect of emotional expressions on patients’ all non-transactional engagement behaviors, whereas professional title weakens their influence on high-effort engagement.
This research enriches our understanding of user behaviors in digital healthcare services by clarifying how dynamic signals and static signals interact to shape non-transactional engagement in high-risk service environments. The findings offer actionable insights for platform algorithm design, physician communication strategies, and the development of trust-oriented online healthcare models.
