Users' Interaction and Social Supports in Online Health Communities about Depression


Student thesis: Doctoral Thesis

View graph of relations



Awarding Institution
Award date13 Sep 2018


Depression among people become more and more serious with the increasing pressure of modern society. It needs long term treatment and is also common among chronic disease patients. In the treatment of depression, social support is important and beneficial to patients’ physical and mental health. Online health communities (OHCs) present a platform on which patients can exchange information and obtain social support from each other. OHCs play an important role in providing online social support, because of their anonymity and ability to break through the patients’ geographical limitations. OHCs provide a new channel that facilitates social support among patients who do not know each other in person. Therefore, to explore users’ interaction and communication could contribute to constructing, maintaining and understanding OHCs well and help improve users’ health condition. In this dissertation, we aim to study the interaction and social supports in OHCs about depression.

First of all, leveraging a unique dataset collected from a large depression OHC in China, we adopted the texting mining method such as topic model and Linguistic Inquiry and Word Count to explore the linguistic properties of the discussion among patients. We found that the initial posts, initiators’ replies and repliers’ posts have different linguistic properties. After receiving the replies, initiators would become more positive, less negative, and less anxious, sad and desperate. In addition, we explored the categrization of social supports based on stress buffing theory and classified them into emotional support, esteem support, network support, information support and other support using Long Short-Term Memory sequence classification. This method showed better performance in OHC text classification problem

Further, we investigated the mechanism of how emotional support in OHC helps patients with depression. Drawn from social support theory, emotional support theory, and social capital theory, we built a framework on support seekers’ emotion considering his or her received replies and the social roles of supporters. Leveraging a unique OHC dataset, we combined text mining, social network analysis, and econometric analysis, and found that the emotion of depressed patients is positively affected by the emotion expressed by the replies they received from other OHC members. The effectiveness of this emotional support is higher if the responses are provided by more active members in the community, or those who have prior interactions with the support seeker, but lower if the responses are provided by more centrally connected members. These findings enrich our theoretical understanding of the emotional contagion in OHCs and demonstrate the value of OHC social support activities for patients with depression, which can be used to direct online interventions for users with depression and improve social welfare.

Finally, we explored the effect of other types of supports in OHCs on users’ health state improvement. We proposed matching embedded hidden Markov model based on the match theory to explore the impact of emotional support, esteem support, network support information support and other support. The results showed that matched social supports could significantly improve patients’ health state. In addition, esteem support has the largest probability to match help seekers’ demand. The health state of patients would gradually decrease if they do not receive matched social supports. These findings demonstrate the significant values of these supports for help seekers, which could guide social workers to provide specific service for those depressed patients.