Termviewer - A Web Application for Streamlined Human Phenotype Ontology (HPO) Tagging and Document Annotation

Area of Science:

• Biomedical Informatics
• Computational Biology
• Clinical Data Science

Background:

• Electronic Health Records (EHRs) contain valuable clinical phenotype data.
• Natural Language Processing (NLP) algorithms can annotate Human Phenotype Ontology (HPO) terms from clinical notes.
• Existing NLP tools may produce erroneous annotations, necessitating manual review.

Purpose of the Study:

• To develop TermViewer, a web application for multi-party collaborative annotation and quality assessment of NLP-processed clinical notes.
• To facilitate the classification of high-quality notes and correction of inaccurate HPO term tagging.
• To enable collaborative manual review of computationally generated HPO annotations.

Main Methods:

• Developed TermViewer, a web application for collaborative annotation of clinical notes.
• Integrated MetaMap and cTAKES, two prominent NLP tools, to identify agreement and disagreement in HPO term tagging.
• Enabled users to view clinical notes with highlighted HPO terms for review and correction.

Main Results:

• TermViewer allows efficient, multi-party review and correction of HPO annotations from clinical notes.
• The application highlights discrepancies between MetaMap and cTAKES, aiding expert review.
• TermViewer supports the creation of high-quality, curated HPO datasets from clinical text.

Conclusions:

• TermViewer enhances the accuracy and reliability of HPO term extraction from clinical notes.
• The tool can be used independently or integrated into machine learning pipelines.
• Facilitates the use of clinical phenotype data for disease research and understanding.