Abstract
BACKGROUND: With the development of online health care, an increasing number of patients are consulting and exchanging social support through online health communities. People with different diseases have varying needs for information and emotional support. However, comparisons of similarities and differences in behavioral patterns among patients with different disease types and their social support needs require further exploration. OBJECTIVE: Using a large-scale dataset of user-generated posts, we aimed to systematically examine how disease type (acute vs chronic) influences the behavioral patterns, emotional expressions, and support-seeking needs of users in online health communities, providing actionable insights for tailored community interventions. METHODS: We identified patients with acute diseases and those with chronic diseases and then crawled corresponding user profiles and post data from the chronic disease online health community (CDOHC) and acute disease online health community (ADOHC). Using a pretrained model, we classified and described the social support performance of users. Subsequently, we conducted a comparative analysis of user behaviors, emotions, and needs by mining behavior patterns and textual content from posts. We performed further social network analysis using user profiles. RESULTS: We identified 492,495 posts from 53,245 users in the CDOHC and 52,047 posts from 23,659 users in the ADOHC. Seeking and providing emotional support were higher in the CDOHC (83,231/492,495, 16.9% and 101,453/492,495, 20.6%, respectively), while seeking and providing information support were higher in the ADOHC (33,993/492,495, 22.8% and 61,128/492,495, 41.0%, respectively). These findings indicate that users with chronic diseases have a higher need for emotional support, while most users with acute diseases want to seek information support. The word co-occurrence network revealed distinct thematic patterns between the 2 communities. In the CDOHC, disease management clusters (8/17, 47%) and emotional clusters (7/17, 41%) showed balanced proportions, reflecting the dual needs of patients with chronic diseases. In contrast, in the ADOHC, posts were overwhelmingly focused on treatment (25/28, 89%), with minimal emotional vocabulary clusters (2/28, 7%). Social network analysis further highlighted these differences. The CDOHC showed the highest edge density in the seeking emotional support subnetwork and reciprocal interactions in 68.0% (83,025/122,095) of providing emotional support connections, indicating robust emotional support exchanges. Meanwhile, the ADOHC exhibited significantly faster post velocity in treatment discussions, consistent with its acute care context. These structural differences aligned with user behavior patterns. Users with chronic diseases maintained strong community bonds (averaging 8.2 connections/user), while users with acute diseases prioritized time-sensitive information (12,823/13,938, 92.0% of queries were related to treatment). CONCLUSIONS: This study contributes to a comprehensive understanding of how disease type influences the social behaviors and emotional expressions of users. The findings provide practical implications for doctors, patients' families, and health care participants regarding targeted support strategies for patients with acute and chronic diseases.