Limitations in ambulation and mobility are common among individuals. However, recognizing information about mobility from clinical data is limited by low coverage in existing terminologies. We extract a seed dictionary from a clinical dataset annotated for mobility information, and expand this dictionary using the VerbNet lexicon and word embeddings. Our combined dictionary achieves high coverage of mobility information, and both expansion methods contribute complementary terms. Future work will expand to alignment with other terminologies.

Learning Objective: Understand data-driven methods for expanding a new terminology for mobility activities

Authors:

Ayah Zirikly (Presenter)
National Institutes of Health

Denis Newman-Griffis, National Institutes of Health
Julia Porcino, National Institutes of Health
Pei-Shu Ho, National Institutes of Health
Jonathan Camacho Maldonado, National Institutes of Health
Maryanne Sacco, National Institutes of Health
Alex Marr, National Institutes of Health

Presentation Materials:

• Abstract/Manuscript - 1