cohd

is a Data Source.

The Columbia Open Health Data (COHD) API provides access to observed clinical frequencies and co-occurrence frequencies from electronic health records at Columbia University Medical Center. The database contains counts and frequencies of conditions, procedures, drug exposures, and patient demographics from the OHDSI common data model, along with statistical associations between clinical concepts. To protect patient privacy, all concepts where count ≤10 were excluded and counts were randomized using Poisson distribution. COHD offers multiple datasets including 5-year (2013-2017) and lifetime data, both in hierarchical and non-hierarchical forms, plus beta temporal co-occurrence data.

Domains

clinical, health, public health, precision medicine

License

Warning: No license entered

Homepage

cohd

Repository

GitHub

Infores ID

infores:cohd

FAIRsharing ID

Unknown

Product Summary

Products

From this Resource
ID Name URL Category Format Description
cohd.api COHD API api ProgrammingInterface http RESTful API providing programmatic ac...
cohd.portal COHD Web Interface cohd.io GraphicalInterface http Interactive web interface for explori...
cohd.notebooks COHD API Examples COHD_API_Example.ipynb (2.7 MB) ProcessProduct http Python Jupyter notebooks demonstratin...
cohd.docs ⚠ COHD Documentation api DocumentationProduct http API documentation covering endpoint d...
From other Resources
ID Name URL Category Format Description
translator.cohd.graph Translator COHD KGX Graph latest GraphProduct kgx-jsonl KGX JSONL graph package for Columbia ...
translator.translator_kg.graph Translator Aggregate KGX Graph latest GraphProduct kgx-jsonl Aggregated KGX JSONL graph package co...

Details

Columbia Open Health Data (COHD)

Overview

The Columbia Open Health Data (COHD) project provides an API for accessing observed clinical frequencies and co-occurrence patterns from electronic health records at Columbia University Medical Center. The database is built on the OHDSI (Observational Health Data Sciences and Informatics) common data model and provides valuable real-world evidence about the prevalence of medical conditions, procedures, and drug exposures in clinical practice.

Data Content

Available Datasets

1. 5-Year Non-Hierarchical Dataset (2013-2017)

  • Clinical data from a 5-year window
  • Counts at the specified concept level only
  • Better underlying data consistency

2. Lifetime Non-Hierarchical Dataset

  • Clinical data from all available dates
  • Larger patient population
  • Broader concept coverage

3. 5-Year Hierarchical Dataset (2013-2017)

  • Counts include patients from descendant concepts
  • Example: Ibuprofen includes all specific formulations (600mg tablets, 400mg tablets, oral suspensions, etc.)
  • Clinical data from 2013-2017

4. Temporal Co-occurrence Data (BETA)

  • Time-aware co-occurrence patterns
  • Still under active development

Data Elements

COHD provides access to:

  • Concept Counts: Number of patients associated with each clinical concept
  • Concept Frequencies: Prevalence (count / total patients) in EHR
  • Co-occurrence Counts: Number of patients with pairs of concepts
  • Co-occurrence Frequencies: Joint prevalence of concept pairs
  • Statistical Associations: Chi-square, relative frequency, and observed-to-expected ratios

Vocabulary Standards

All clinical concepts are coded using standard concept IDs from the OMOP Common Data Model:

  • Conditions (e.g., SNOMED CT)
  • Procedures (e.g., CPT, HCPCS)
  • Drugs (e.g., RxNorm)
  • Demographics

The API includes mapping functionality to translate between vocabularies and other ontologies using the EMBL-EBI Ontology Xref Service (OxO).

Privacy Protection

To protect patient privacy:

  • All concepts and concept pairs with count ≤10 are excluded
  • Counts are randomized using Poisson distribution
  • No patient-level data is exposed
  • Aggregate statistics only

API Resources

1. Metadata

Dataset descriptions, concept counts, and database statistics

2. OMOP Vocabulary

Common vocabulary for name and concept identifier mapping between standard terminologies

3. Clinical Frequencies

Access to counts and frequencies of:

  • Conditions
  • Procedures
  • Drug exposures
  • Patient demographics

4. Concept Associations

Inferred associations between concepts using:

  • Chi-square analysis: Statistical significance of association
  • Relative frequency: Ratio of observed to expected co-occurrence
  • Observed-to-expected ratio: Strength of association

Use Cases

  1. Hypothesis Generation: Discovering potential drug-disease associations
  2. Comorbidity Analysis: Understanding disease co-occurrence patterns
  3. Drug Safety: Identifying adverse event signals
  4. Clinical Decision Support: Providing EHR-based prevalence estimates
  5. Phenotype Definition: Understanding clinical manifestations
  6. Drug Repurposing: Finding new therapeutic applications
  7. Resource Annotation: Enriching knowledge graphs with EHR frequencies

Development Team

COHD was developed at the Columbia University Department of Biomedical Informatics through collaboration between:

  • Weng Lab: Chunhua Weng, PhD
  • Tatonetti Lab: Nicholas Tatonetti, PhD
  • NCATS Biomedical Data Translator Program (Red Team)

Funding Support

  • NCATS OT3TR002027
  • NLM R01LM009886-08A1
  • NIGMS R01GM107145

External Resources

Information Resource ID

This resource has the Information Resource identifier: infores:cohd

Technical Details

  • Data Source: Columbia University Medical Center EHR
  • Data Model: OHDSI OMOP CDM
  • Update Frequency: Static snapshots from specified time periods
  • API Format: REST with JSON responses
  • Authentication: No authentication required

Is this information incorrect or incomplete? Request an update.

Created: November 04, 2025 | Last modified: November 04, 2025