AI Unlocks Eating Disorder Insights on Chinese Social Media

Eating disorders (EDs) represent a significant public health challenge, marked by high mortality rates and severe medical complications. In China, this challenge is compounded by low rates of diagnosis and treatment engagement, leading to delayed interventions. While social media platforms offer a rich, naturalistic window into individuals' experiences with EDs, there has been a notable scarcity of research focusing on Chinese-language data.

To address this gap, a pioneering study utilized advanced machine learning (ML) and deep learning (DL) techniques to systematically identify and characterize ED-related discourse on Weibo, a prominent Chinese social media platform. The researchers collected posts through keyword-based API searches, then meticulously categorized them into irrelevant, promotional/educational, and layperson content. A two-stage classification framework was developed, employing five different ML/DL methods, including Convolutional Neural Networks (CNNs), to first filter out irrelevant posts and then differentiate between promotional and layperson content.

The study found that Convolutional Neural Networks (CNNs) consistently outperformed other models, achieving impressive F1-scores of 0.87 and 0.98 in the two classification stages, respectively. Following successful classification, Latent Dirichlet Allocation (LDA) was applied to the layperson posts to uncover underlying themes. This topic modeling revealed five critical themes: restrictive symptomatology and physical distress, binge eating and body-health concerns, relapse and coping narratives, emotional venting, and chronic ED patterns with identity impact.

These findings not only represent a significant methodological advance in non-English Natural Language Processing but also provide novel, culturally specific insights into how eating disorders are experienced and discussed within the Chinese context. The demonstrated CNN-based classification combined with topic modeling offers a scalable framework for public health surveillance. As lead author Zhang notes in the paper, "These insights can inform the development of early detection tools and culturally sensitive interventions to address the unmet needs of individuals with EDs in China."