TermViewer

TermViewer facilitates collaborative annotation and quality assessment of clinical notes using Human Phenotype Ontology (HPO) terms to improve phenotype extraction from electronic health records for research and clinical applications.

Key Features:

• Multi-Party Collaborative Annotation: Enables multiple annotators to assign HPO terms to clinical notes to increase annotation accuracy and reliability.
• Quality Assessment: Supports assessment and improvement of annotations produced by Natural Language Processing (NLP) algorithms.
• Integration with NLP Tools: Processes clinical notes with MetaMap and cTAKES and highlights areas of agreement and disagreement between their outputs.
• Highlighting HPO Terms: Detects and highlights HPO terms within clinical text to guide correction of incorrect annotations.
• Stand-Alone and Pipeline Integration: Operates stand-alone or exports tagged HPO terms as input for machine-learning pipelines to support model training.

Scientific Applications:

• Phenotype extraction from EHRs: Produces HPO-tagged phenotype datasets from electronic health record clinical notes for downstream analyses.
• Disease mechanism research: Provides improved phenotype datasets to support studies of disease mechanisms.
• Patient stratification: Facilitates standardized phenotype-based cohort selection and subgroup analyses.
• Genotype–phenotype and personalized medicine: Improves phenotype inputs for genotype–phenotype analyses and personalized medicine approaches.
• Large-scale annotation workflows: Supports multi-expert annotation workflows for scalable phenotype curation across studies.

Methodology:

Clinical notes are processed through MetaMap and cTAKES to generate initial HPO annotations and to identify concordant and discordant term matches; tagged HPO terms can be exported for use in machine-learning pipelines.