This study explored extracting scientific evidence from published abstracts about the effectiveness of various physical activity interventions provided to NAFLD patients. We focused on extracting interventions, effectiveness, and clinical parameters that an intervention was targeted for. We used CLAMP to annotate 108 PubMed abstracts. The results indicated our text mining methods require further improvement to accurately assess the feasibility of systematically extracting and profiling the physical activity interventions and their effects in managing NAFLD.
Learning Objective: 1. The audience will understand the challenges associated with extracting information from published abstracts
2. The audience will understand the limitations in "naive" NLP approaches to extracting information from published abstracts
William Gu, Duke University
Yujing Ke, Duke University
Shenghong Zhao, Duke University
Hyeoneui Kim (Presenter)