Skip to main content

025

Title:#

Curating Interface Ontologies by Extracting and Integrating Concepts from Heterogeneous Sources

Discipline: Computer Science

Presenter:#

Vipina Keloth

Abstract:#

During the past 10 years, the federal government has taken the initiative to encourage the broad use of Electronic Health Records (EHRs). EHRs have great potential to improve the health care system by providing accurate, up-to-date information about patients, enabling quick access to patient records and enabling secondary use of data, e.g., for epidemiological studies. EHRs should facilitate ease of use and interoperability between providers, and support medical research. Unfortunately, recent studies have exposed EHR data entry as one of the reasons for physician burnout and clinician stress. Some US physicians now spend more time “treating the EHRs” than they do face-to-face with patients. Existing EHR patient data entry is cumbersome and healthcare providers face difficulties in finding the exact medical terms to enter into the records of their patients. In this project, we propose to create interface ontologies, i.e., well organized collections of medical terms, that are specifically designed to ease this difficulty for a specific medical specialty, e.g., cardiology. We are creating a Cardiology Interface Ontology by gathering heart-related concepts from existing ontologies and by mining EHR text notes to extract concepts related to cardiology. We hypothesize that our interface ontology will have broader, better-organized coverage than existing ontologies, facilitating ease of use. For developing interface ontologies, we deploy several techniques, namely structural comparison of identical concepts in existing medical terminologies, existing annotator programs, and recent AI Machine Learning language models for concept extraction from EHRs.

Author(s):#

Vipina Keloth, James Geller, Yehoshua Perl

Funding Acknowledgements:#

Parts of this project was supported by the National Cancer Institute of the National Institutes of Health under Award Number R01CA190779.