**4. Discussion**

Twitter enables millions of users to share information worldwide in real time. This phenomenon allows policymakers, healthcare stakeholders, and other people to influence and be influenced by opinions and discussions that flow across online social communities, making it possible to share valuable information and practices more quickly and easily than ever before. Such a possibility has become a rich source of value for information-gathering and practical dissemination, in particular for complex and low-intensity diseases like Lupus [46].

Interactivity among online communities makes it possible to renovate not only the healthcare organizations' online approach but the manner in which people's attitudes and intentions regarding health behavior might be influenced [47]. However, valuable information is complex to detect and depict, considering how vast and fast social media platforms work, too often spreading rumors or misinformation [48].

For these reasons, investigating the dynamics played by online communities during specific events like the World Lupus Day can offer a powerful tool to stakeholders for identifying and setting up policy strategies for gathering valuable information and sharing good practices. This ability can offer a concrete tool for decision-makers in dealing with information asymmetry [49], obtain valuable new elements for the decision-making process, promote trust across the identified communities, and promote health-seeking behavior [50].

In our study, we sought to analyze latent themes spread on Twitter during World Lupus Day and detect online user communities' behavior by measuring the users' retweet network.

We measured and found two different types of influencers in our analysis, who behaved and acted differently. There was one type of influencer who was more generally public-oriented, measured on the ratio between the number of followers and the ability to amplify the content they posted, and a second type of influencer, more based on the retweets and network attention count, as an endorsement of their tweet content.

Network influencer users, mostly led by patient organizations, have many followers who tend to have intense connections among themselves, and show more interest in specific topic discussions about the role of social support and policy advocacy. General influencers show less network connection and appear to attract more followers with content related to general disease advice.

Many topics discussed by the two types of influencer were in common. However, the attention posed in some topics were different. This is well represented by the discussion order of topics 2 (Information and advice), 5 (social support), and 6 (advocating), which are swapped in the likelihood order.

Another difference between the topics was posed by the fact that general influencers discussed body symptoms (topic 11), whereas the network influencer discussed topics related to patient involvement (topic 9) and diseases description (topic 8).

To the best of our knowledge, this study is the first to employ a combination of methods to explore deeply latent topic discussions and online communities' interactions regarding a low-prevalence disease like Lupus. Unlike other kinds of diseases such as diabetes, HIV, or stroke, where the vast population offers more opportunities for investigation, low-prevalence or rare diseases can benefit greatly from the application of such methodologies to identify unmet needs or improve the network of care and treatment for patient communities. Therefore, it is critical for public health institutions to systematically explore how to effectively use interactive features on social media to attract public attention and maintain communication with the public.

Further research should also evaluate a qualitative analysis of the selected topics, offering insights that can help improve the judgement in understanding the topic relationships [51].
