Different platforms for different patients’ needs: Automatic content analysis of different online health information platforms

• Published in: International journal of human-computer studies 2020-05, Vol.137, p.102386, Article 102386
• Main Authors: Sanders, Remco, Linn, Annemiek J., Araujo, Theo B., Vliegenthart, Rens, van Eenbergen, Mies C., van Weert, Julia C.M.
• Format: Article
• Language: English
• Subjects: Automatic content analysis; Cancer; Information needs; Information seeking behavior; Model of patients’ social support needs; Online health information
• ISSN: 1071-5819
• DOI: 10.1016/j.ijhcs.2019.102386

•Peer-generated platforms mostly contain content on affective support needs whereas expert-generated platforms mostly contain content on cognitive support needs.•Coupling topic modeling and a-priori defined theoretical models in a hybrid method proof to be useful in analyzing online health information data.•In line with the optimal matching model, patients should be referred to, or seek out, specific platforms depending on their needs. Prior online health research has mainly focused on the predictors or outcomes of online health information, leaving online health information itself understudied. Therefore, online health information has remained an umbrella term encompassing different platforms (expert- vs. peer-generated). A hybrid method that combines qualitative and computational methods is used to identify different topics discussed on these different platforms, and an initial model of patients’ social support needs was developed and applied to data obtained from the hybrid method. Using topic modeling (Nposts = 52.990), topics on two expert- and two peer-generated platforms were identified. Differences between and within platforms were found. While peer-generated platforms mainly covered interaction on emotional support topics, expert-generated platforms covered informational topics. Within peer-generated platforms, patients used their experiences differently.