Computational analysis of neurodevelopmental phenotypes: Harmonization empowers clinical discovery

Area of Science:

• Genomics
• Clinical Informatics
• Neurodevelopmental Disorders

Background:

• Traditional neurodevelopmental disorder diagnosis relies on recognizing distinct syndromes to guide genetic testing.
• Genomic diagnostics face challenges integrating meaningful clinical phenotypic measurements.
• Standardization of genomic data has advanced, but clinical data standardization lags.

Purpose of the Study:

• To develop novel computational phenotyping approaches for harmonizing and quantifying clinical features.
• To improve data translation and clinical relatedness assessment in genomic diagnostics.
• To leverage longitudinal and electronic medical record data for enhanced diagnostic insights.

Main Methods:

• Harmonizing clinical features using controlled vocabularies like the Human Phenotype Ontology (HPO).
• Revising domain-specific dictionaries to improve data translation.
• Quantifying phenotypic features and assessing clinical relatedness computationally.
• Applying methods to longitudinal data and electronic medical records.

Main Results:

• Demonstrated novel approaches for computational phenotyping.
• Showcased application to longitudinal phenotypic information crucial for developmental disorders.
• Highlighted the potential of electronic medical records as a rich data source.

Conclusions:

• Computational phenotyping offers a pathway to integrate clinical data effectively into genomic diagnostics.
• Standardized and quantifiable clinical data can significantly inform genomic research and diagnosis.
• Leveraging diverse clinical data sources, including EMRs, is key for advancing diagnostic capabilities.