**5. Limitations**

For all its strengths, this study has limitations. We based our analysis on just one specific day that may not describe all the dimensions and themes about Lupus Awareness Month. Data collection relied on a public Twitter API was able to detect 4434 tweets in English, which may have led to a loss of some tweets.

In the dataset, most accounts were based in the UK or USA due to the language choice. Only a tiny percentage of accounts reported the geographical location, making it impossible to properly explore specific geographic characteristics at the country level. Therefore, future studies could take into account and explore a longer period, consider other languages, and evaluate geographic and ethnic effects that play a role in Lupus.

We used structural topic modeling to analyze tweet texts, while other methods may offer other types of classification based on natural processing language or deep learning suitable for tweet texts [52]. However, despite these limitations, this study provides an extensive and detailed methodological approach offering useful insights into social media platform dynamics regarding Lupus, which is still little investigated.
