Eating disorders, such as anorexia and bulimia, are complex mental illnesses that can have serious consequences for an individual's physical well-being, in addition to having the highest mortality rate among all mental illnesses. Given the stigma that currently persists around these issues, affected individuals often conceal their symptoms, and many seek social support or information in online communities, especially through social media platforms like Twitter.
With the presence of these interactions on social media, studying them can enhance our understanding of the dissemination and integration of these issues in other communities. Despite not commonly sharing a direct relationship, the communities chosen for this study (fitness, mental health, depression, and self-harm) were considered relevant due to the following characteristics: obsessive focus, emotional issues, potential unhealthy relationship with food and body, shame, and associated guilt. To achieve this, we collected a broad set of Twitter users based on a preselection of hashtags with a higher probability of intersection, aiming to capture direct conversations between individuals through "reply" and "mention" interactions.
We represented these interactions using an undirected and weighted network, measuring network structures to reveal how users interact with each other and the potential communities formed, without considering the root community of the interaction.
The Python code can be found in the /code directory, and the generated NetworkX graphs are located in the /graphs section of this repository.
For more information about the study, check out my blog post ! :)