JMIR Publications

ABSTRACT

Background:

In recent years, an increasing number of users join online health communities (OHCs) to obtain information and seek for support. In particular, patients often look for suggestions and information to support them make decisions in their diagnosis and treatments. It is important to understand patient decision-making processes and identify the influences patients receive from OHC.

Objective:

We aim to identify the posts in discussion threads that made influences on users who seek for help in their decision-making.

Methods:

We proposed the definition of influence relationship of posts in discussion threads. Then we developed a framework and a deep learning model for identifying influence relationships. We leveraged the state-of-the-art text relevance measurement methods to generate sparse feature vectors to present the text relevance. We modeled the probability of question and action presence in a post as dense features. Then we use deep learning techniques to combine the sparse and dense features to learn influence relationships.

Results:

We evaluated the proposed techniques on a large-scale dataset of discussion threads from a popular cancer survivor OHC. Empirical evaluation demonstrated the effectiveness of our approach.

Conclusions:

It is feasible to identify influence relationships in discussion forums. Using the proposed techniques, a significant amount of discussions on OHC are identified to have made influences. Such discussions are more likely to affect user decision-making processes and to engage users’ participation at OHC. Studies on those discussions can help to improve information quality, user engagement and user experience.