eHDPrep

eHDPrep prepares and semantically enriches electronic health datasets to improve data quality and machine interpretability for downstream analyses.

Key Features:

• Quality Control: Performs internal consistency checks, redundancy removal, and information-theoretic variable merging to streamline datasets without discarding critical information.
• Semantic Enrichment: Leverages ontologies including SNOMED CT and the Gene Ontology to identify ontological relationships and generate meta-variables derived from the common ancestry of input variables.
• Data Preparation: Implements numerical encoding of categorical variables, extraction of variables from free-text, and completeness analysis to assess missing data.
• Modification Tracking: Records modifications made during dataset preparation to enable review and traceability of processing steps.

Scientific Applications:

• Multi-modal health research: Supports preprocessing and semantic enrichment of clinical and genomic data to create structured datasets for downstream analyses.
• Colorectal cancer dataset analysis: Enhances data quality, structuring, and semantic content in colorectal cancer datasets to support analytical workflows.

Methodology:

Implements internal consistency checks, redundancy removal, information-theoretic merging of variables, ontological integration with SNOMED CT and the Gene Ontology to derive meta-variables via common ancestry, numerical encoding, free-text variable extraction, and completeness assessment.