Social determinants of health (SDOH) negatively impact morbidity and mortality especially in marginalized and vulnerable populations. For clinical care, it has long been practice to record SDOH as part of social history during clinical care as this history impacts not only diagnosis but also treatment options. SDOH information can improve public health through secondary use applications, like clinical decision-support systems; however, SDOH must first be automatically extracted from unstructured clinical narrative text. In this work, we apply automatic SDOH extraction methods to two publicly available datasets, namely i2b2 Obesity Challenge dataset [1] and MIMIC-3 dataset [2], and present prevalence of SDOH information in these corpora.

Learning Objective: Application of NLP for extraction social determinants of health

Authors:

Kevin Lybarger, University of Washington
Mari Ostendorf, University of Washington
Ozlem Uzuner, George Mason University
Meliha Yetisgen (Presenter)
University of Washington

Presentation Materials:

• Abstract/Manuscript - 1