The aim of the Social Media Mining for Health (#SMM4H) shared tasks is to take a community-driven approach for developing and evaluating natural language processing, machine learning, and artificial intelligence methods to utilize publicly available social media data for health research. For the 10th iteration, hosted at the AAAI International Conference on Web and Social Media (ICWSM) 2025, we broadened the scope to include additional web-based sources of "health real-world data" (HeaRD). The 6 tasks representedvarious data sources (Twitter, Reddit, patient forums, clinical notes, news articles), languages (English, Russian, German, French), health-related topics (adverse drug and vaccine events, nonmedical substance use, dementia family caregiving, insomnia, foodborne disease outbreaks), and methods (binary classification, multi-class classification, named entity recognition). In total, 57 teams registered, representing 17 countries. In this paper, we present an overview of the annotated corpora, participants' systems, and performance results, providing insights into state-of-the-art methods for mining social media and other web-based data sources for health research. To facilitate future work, the datasets remain available by request, and the CodaLab sites remain active for a post-evaluation phase.

PDF