Get Started

Welcome to the OHDSI Waveform Working Group! This page helps you get started whether you’re implementing the Waveform Extension, joining the working group, or proposing a use case.


Quick Start: Understanding the Waveform Extension

The OMOP CDM Waveform Extension enables standardized integration of physiological waveform data with electronic health records. The extension consists of four interconnected tables:

Implementation Overview

1. waveform_occurrence    → Define the recording session context
2. waveform_registry      → Register individual waveform files
3. waveform_channel_metadata → Describe signal-level metadata
4. waveform_feature       → Store derived features and measurements

Prerequisites: - OMOP CDM v5.4+ database - Physiological waveform data (ECG, EEG, ABP, etc.) - ETL development environment - Basic understanding of signal processing (recommended)

What you’ll need: - Raw waveform files with timestamps - Patient identifiers linkable to PERSON table - Visit/encounter context for acquisitions - Device and procedure metadata (if available)

See the Waveform Extension Implementation Guide for detailed ETL specifications.


Join the Working Group

Get Connected

  1. Sign up for OHDSI Teams

  2. Join the Waveform Working Group

    • Request access to the Waveform WG Teams channel
    • Introduce yourself and share your interests

  3. Attend Working Group Meetings

    • Format: Use case presentations, technical discussions, implementation Q&A
    • Contact: for meeting schedule

  4. Introduce Yourself

    • Share your background (clinical, data engineering, research, etc.)
    • Describe what types of waveform data you work with
    • Explain what you hope to accomplish with the extension

Start Contributing

For Data Engineers: - Review the Implementation Guide - Check Table Specifications - Share your ETL approach and challenges - Contribute reference implementations to GitHub

For Researchers: - Propose use cases that require waveform data integration - Identify required features not yet in the specification - Share validation requirements - Collaborate on multi-site studies

For Developers: - Build ETL tools for common waveform formats (EDF, WFDB, etc.) - Develop visualization tools - Create data quality checks - Contribute to vocabulary development

For Vocabulary Experts: - Map waveform concepts to OMOP standard concepts - Standardize device and procedure terminology - Extend vocabularies for signal types and features


Implementation Steps

Phase 1: Planning

  1. Assess Your Data
    • Identify waveform data sources (file formats, storage locations)
    • Determine linkage keys to OMOP PERSON and VISIT_OCCURRENCE
    • Document device types and acquisition contexts
    • Estimate data volumes and storage requirements
  2. Define Scope
    • Which waveform types will you include? (ECG, EEG, ABP, etc.)
    • Will you ingest historical data or only new acquisitions?
    • What derived features do you need? (HR, HRV, respiratory rate, etc.)
    • Do you need real-time ingestion or batch processing?
  3. Set Up Infrastructure
    • Extend OMOP database schema with waveform tables
    • Plan file storage strategy (object storage, file system, etc.)
    • Set up ETL development environment
    • Prepare test datasets

Phase 2: Implementation

Follow the Waveform Extension Implementation Guide to:

  1. Populate waveform_occurrence
    • Extract recording session metadata
    • Link to patients and visits
    • Establish temporal boundaries
  2. Populate waveform_registry
    • Register each waveform file
    • Store source and target file paths
    • Record file-level temporal information
  3. Populate waveform_channel_metadata
    • Extract per-signal metadata (sampling rate, gain, units)
    • Map channel names to standard concepts
    • Link to device and procedure records
  4. Populate waveform_feature (optional)
    • Apply feature extraction algorithms
    • Store derived measurements
    • Link to MEASUREMENT/OBSERVATION tables

Phase 3: Validation

  1. Data Quality Checks
    • Verify referential integrity across tables
    • Check temporal consistency
    • Validate concept mappings
    • Test file accessibility
  2. Test Queries
    • Retrieve waveforms for specific patients/visits
    • Join waveform features with clinical events
    • Aggregate features across populations
    • Test performance at scale
  3. Share Results
    • Document implementation patterns
    • Report data quality metrics
    • Share ETL code (if possible)
    • Provide feedback to working group


Common Use Cases

ICU Telemetry Integration

Scenario: Integrate continuous bedside monitor data (ECG, ABP, SpO2) with ICU EHR data

Implementation: - Link telemetry sessions to ICU visits via timestamps - Register multi-channel waveform files (often EDF or proprietary formats) - Extract vital signs as features (HR, BP, RR, SpO2) - Align with medication administration and interventions

Applications: - Early warning systems for hemodynamic instability - Medication effect analysis - ICU length of stay prediction

Diagnostic ECG Studies

Scenario: Standardize 12-lead diagnostic ECG data from cardiology clinics

Implementation: - Link ECGs to outpatient visits or procedures - Register ECG files (typically HL7 aECG, DICOM, or MUSE XML) - Extract automated measurements (QT, QRS, PR intervals) - Map ECG interpretations to OMOP concepts

Applications: - QT prolongation risk assessment - Arrhythmia phenotyping - Medication safety surveillance

EEG Analysis

Scenario: Enable neurological research with EEG and clinical data

Implementation: - Link EEG studies to neurology encounters - Register multi-channel EEG files (often EDF+) - Extract power spectral features, entropy, or seizure annotations - Align with medication, imaging, and outcome data

Applications: - Seizure prediction models - Sleep disorder research - Post-operative cognitive monitoring


Resources

Documentation

Repositories

  • OHDSI/WaveformWG - Central working group repository
  • Example ETL code (coming soon)
  • Vocabulary mappings (coming soon)

Getting Help

Questions?

Technical Support

New to Waveforms/OMOP/OHDSI?

Welcome! We support all skill levels: - Pair with experienced implementers - Start by reviewing existing implementations - Ask questions freely during meetings - Review the OHDSI Community resources - Explore PhysioNet datasets for learning


Contact