Natural Language Processing Pitfalls in Patient Phenotype Extraction

Poster Number: P38
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Terminology Systems, Standards
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Whole Exome Sequencing (WES) is a powerful genomic technique that links genetic variations to patient phenotypes to support precision medicine. However, manual abstraction of patient phenotypes for providers is labor-intensive and lack integration with Electronic Health Records (EHR). Our study evaluated an enterprise Natural Language Processing (NLP) application to extract standard Human Phenotype Ontology (HPO) concepts, revealing “semantic” challenges in NLP detection process . Our findings underscore the need for NLP optimization and standardization in genomic workflows to enhance data accuracy and streamline structured data capture and usage processes.

Speaker(s):
Sina Madani, MD, PhD
Vanderbilt University Medical Center

Author(s):
Sina Madani, MD, PhD - Vanderbilt University Medical Center; Asli Ozdas Weitkamp, PhD - Vanderbilt University Medical Center; Dario Giuse, DrIng - Vanderbilt University Medical Center;