Harvard Electroencephalography Database

Harvard Electroencephalography Database aggregates over 280,000 de-identified, standardized electroencephalographic recordings from four Harvard-affiliated hospitals to support artificial intelligence-driven and reproducible research in epilepsy and clinical neuroscience.

Key Features:

• Large-Scale Data Aggregation: HEEDB contains more than 280,000 EEG recordings from more than 108,000 patients across four Harvard-affiliated hospitals.
• Data Harmonization: Data are harmonized using the Brain Imaging Data Structure and hosted on the Brain Data Science Platform.
• Multimodal Data Linkage: EEG data are linked with clinical notes, International Classification of Diseases, 10th Revision codes, medications, and EEG reports.
• HIPAA-Compliant Deidentification: Deidentification follows Health Insurance Portability and Accountability Act Safe Harbor standards.
• Diverse Recording Types: The database includes routine, epilepsy monitoring unit, and intensive care unit EEGs across all age groups.
• Clinical Report Integration: 73% of recordings are linked to deidentified clinical reports, with 96% of those matched to recordings.
• Automated Finding Extraction: Findings are extracted using expert curation, regular expressions, and medical natural language processing models.
• Auxiliary Clinical Data: Diagnoses, medications, and hospital course data support multimodal analysis.
• FAIR Alignment: The database promotes data-sharing in alignment with FAIR (Findable, Accessible, Interoperable, Reusable) and National Institutes of Health data policies.

Scientific Applications:

• Epilepsy Research: HEEDB fills a critical gap in EEG data availability for epilepsy research.
• Diagnostic Tool Development: The database accelerates the development of diagnostic tools through large-scale, privacy-compliant, and clinically relevant analysis.
• Machine Learning Training: HEEDB improves training datasets for machine learning models in clinical neuroscience.
• Multimodal Clinical Analysis: Linked diagnoses, medications, and hospital course data enable multimodal analysis of neurological conditions.

Methodology:

Data are harmonized using the Brain Imaging Data Structure. Findings are extracted using expert curation, regular expressions, and medical natural language processing models. Deidentification follows Health Insurance Portability and Accountability Act Safe Harbor standards.