This research seeks to explore the discussions surrounding JAK inhibitors on Reddit by utilizing graph modeling and community detection techniques through the application of NetworkX and the Louvain algorithm.
Data was gathered from Reddit by targeting keywords related to JAK Inhibitors. Then cleaned and preprocessed to get rid of any noise and missing information. Also, NetworkX and Louvain algorithm were used to construct graph networks and detect the communities respectively. Our study revealed 288 communities talking about JAK inhibitors and implemented the betweenness centrality measures to detect the key influencers within these groups. Moreover, Large Language Models (LLMs) such as GPT 3.5 model were used to detect misinformation regarding the usage and potential adverse effects of these medications.
The results show that certain Reddit communities often share information actively with certain users acting as key influencers within the groups. This study deepens our understanding of the dynamic conversation surrounding JAK inhibitors by identifying the communities and users propagating this information. It also highlights the efficacy of graph-based models in addressing misinformation within conversation related to health.