RedHOT: A Corpus of Annotated Medical Questions, Experiences, and Claims on Social Media

• DOI: 10.48550/arxiv.2210.06331
• Publication Date: 2022-01-01
• Authors: Somin Wadhwa; Vivek Khetan; Silvio Amir; Byron Wallace
• Keywords: Trustworthiness; Baseline (sea)

We present Reddit Health Online Talk (RedHOT), a corpus of 22,000 richly annotated social media posts from spanning 24 health conditions. Annotations include demarcations spans corresponding to medical claims, personal experiences, and questions. collect additional granular annotations on identified claims. Specifically, we mark snippets that describe patient Populations, Interventions, Outcomes (PIO elements) within these. Using this corpus, introduce the task retrieving trustworthy evidence relevant given claim made media. propose new method automatically derive (noisy) supervision for which use train dense retrieval model; outperforms baseline models. Manual evaluation results performed by doctors indicate while our system performance is promising, there considerable room improvement. Collected (and scripts assemble dataset), are available at https://github.com/sominw/redhot.