Skip to content

BEP: Standardized Medication/Pharmaceutical Metadata for BIDS #2295

New issue
New issue
Open
Open
BEP: Standardized Medication/Pharmaceutical Metadata for BIDS#2295
@yarikoptic

Description

@yarikoptic
yarikoptic
opened on Jan 7, 2026

There was:

which touched upon adding description of medication(s) for a patient for iEEG (attn @bids-standard/raw-electrophys-ieeg @bids-standard/raw-electrophys-eeg @bids-standard/raw-electrophys-meg). It was never addressed, and I do not think it is iEEG specific anyhow, thus I will close that issue in favor of this more generic BEP proposal.

We already have some Pharmaceutical* fields described:

It would be great to formalize, taking into account work in #319.

Motivation

Medication/pharmaceutical information is critically important across many neuroimaging modalities:

  • iEEG/EEG/MEG: Anti-epileptic drugs significantly affect signal quality and spike detection
  • PET: Pharmacological challenges during scans, tracer interactions
  • fMRI: Drug studies, effects on BOLD signal
  • Animal electrophysiology: Anesthetics, experimental compounds

Currently, medication metadata is either absent, inconsistently represented, or limited to specific modalities. A unified approach would improve:

  1. Reproducibility: Clear documentation of pharmacological state during data acquisition
  2. Data sharing: Standardized format enables cross-study comparisons
  3. Analysis: Machine-readable medication data supports automated phenotyping

Proposal Overview

This BEP proposes:

  1. Adoption of established controlled vocabularies/ontologies for medication identification
  2. A standardized schema for medication administration details (dose, route, timing, etc.)
  3. Modality-agnostic placement that can extend current PET Pharmaceuticals fields

Recommended Ontologies for Drug/Medication Identification

RxNorm (NIH/NLM) - Primary recommendation for clinical/US contexts
  • URL: https://www.nlm.nih.gov/research/umls/rxnorm/
  • API: https://lhncbc.nlm.nih.gov/RxNav/APIs/RxNormAPIs.html
  • License: Free, non-proprietary (NLM)
  • Description: The de facto US standard for normalized drug names. Provides normalized names for clinical drugs and links to many drug vocabularies (First Databank, Micromedex, Multum, Gold Standard).
  • Key features:
    • RxCUI (unique concept identifier) for each drug formulation
    • Links ingredients, strengths, dose forms
    • Mappings to NDC codes, ATC classifications
    • Well-maintained REST API
  • Example: Carbamazepine 200 MG Oral Tablet → RxCUI: 308971
ChEBI (Chemical Entities of Biological Interest) - Primary recommendation for FAIR/open science
  • URL: https://www.ebi.ac.uk/chebi/
  • License: Fully open (CC BY 4.0)
  • OBO Foundry member: http://obofoundry.org/ontology/chebi.html
  • Description: Open-access database and ontology of molecular entities from EMBL-EBI. Part of the OBO Foundry family, designed for interoperability with other biomedical ontologies.
  • Key features:
    • CHEBI IDs are stable, persistent identifiers
    • Full chemical structure information
    • Role-based classification (e.g., "anticonvulsant", "anesthetic")
    • Integration with Gene Ontology, Reactome, BioModels
    • Available in OBO, OWL, SDF formats
  • Example: Carbamazepine → CHEBI:3387
  • Relevance: Already used in OMOP2OBO mappings for drug ingredients; aligns with FAIR principles central to BIDS philosophy
ATC (Anatomical Therapeutic Chemical Classification) - WHO standard
  • URL: https://www.who.int/tools/atc-ddd-toolkit
  • License: WHO (free for research)
  • Description: Hierarchical classification system from WHO, widely used internationally in pharmacoepidemiology.
  • Key features:
    • 5-level hierarchy (Anatomical → Therapeutic → Pharmacological → Chemical → Substance)
    • Includes Defined Daily Doses (DDD) - useful for dosing context
    • Standard in clinical trial submissions (via CDISC)
  • Example: Carbamazepine → N03AF01 (N=Nervous system, 03=Antiepileptics, AF=Carboxamide derivatives, 01=Carbamazepine)
DrugBank
  • URL: https://go.drugbank.com/
  • License: Mixed (open + commercial tiers)
  • Description: Comprehensive drug database with detailed pharmacological information including targets, interactions, pharmacokinetics.
  • Key features:
    • Rich pharmacological metadata
    • Drug-target interactions
    • ADMET properties
  • Note: Commercial licensing may limit use in fully open datasets
SNOMED CT (Medications subset)
  • URL: https://www.snomed.org/
  • License: Requires SNOMED license (free in many countries via national release centers)
  • Description: Comprehensive clinical terminology including medications, widely adopted in EHR systems internationally.
  • Note: Licensing complexity may be a barrier for international data sharing

Recommended Schemas for Medication Administration

HL7 FHIR MedicationStatement / MedicationAdministration - Primary recommendation

Key fields from FHIR that could inform BIDS medication metadata:

{
  "medication": {
    "code": {
      "coding": [
        {"system": "http://www.nlm.nih.gov/research/umls/rxnorm", "code": "308971"},
        {"system": "https://www.ebi.ac.uk/chebi", "code": "CHEBI:3387"},
        {"system": "http://www.whocc.no/atc", "code": "N03AF01"}
      ],
      "text": "Carbamazepine 200mg"
    }
  },
  "dosage": {
    "dose": {"value": 200, "unit": "mg"},
    "route": {"text": "oral"},
    "timing": {"repeat": {"frequency": 2, "period": 1, "periodUnit": "d"}}
  },
  "effectivePeriod": {
    "start": "2024-01-01",
    "end": "2024-06-15"
  },
  "status": "active",
  "reasonCode": [{"text": "epilepsy"}]
}

Advantages:

  • Open, JSON-native, widely adopted in healthcare
  • Handles medication coding with multiple vocabulary references
  • Comprehensive dosage model (amount, timing, route, method)
  • Designed for real-world medication documentation scenarios
CDISC SDTM CM (Concomitant Medications) Domain

Key SDTM CM variables:

Variable Description
CMTRT Reported Name of Drug
CMDECOD Standardized Medication Name (dictionary-derived)
CMCLAS Medication Class (e.g., ATC level)
CMCLASCD Medication Class Code
CMDOSE Dose per Administration
CMDOSU Dose Units
CMDOSFRQ Dosing Frequency per Interval
CMROUTE Route of Administration
CMSTDTC Start Date/Time
CMENDTC End Date/Time
CMINDC Indication

Advantages:

  • FDA-accepted standard for clinical trial submissions
  • Well-defined field specifications
  • Established mapping to WHO Drug Dictionary
OpenEHR Medication Archetypes

Related Projects and Standards

OHDSI / OMOP CDM
  • URL: https://www.ohdsi.org/
  • Drug Exposure table: https://ohdsi.github.io/CommonDataModel/cdm60.html#DRUG_EXPOSURE
  • Description: Observational Health Data Sciences and Informatics consortium has done extensive work mapping drug vocabularies for large-scale health data research. Their OMOP2OBO project maps OMOP drug concepts to OBO ontologies including ChEBI.
  • Relevance: Demonstrates successful integration of RxNorm, ATC, and ChEBI for drug data standardization
Biolink Model
schema.org Drug type
  • URL: https://schema.org/Drug
  • Description: Lightweight structured data vocabulary. Limited but provides basic drug metadata fields.

Proposed BIDS Approach

Vocabulary Selection

Primary recommendation: Support multiple vocabularies with clear preference order:

  1. RxNorm (RxCUI) - for US clinical drugs, normalized names
  2. ChEBI (CHEBI ID) - for full FAIR compliance, international use, OBO integration
  3. ATC - for therapeutic classification, DDD information

Datasets SHOULD include at least one of these identifiers. Free-text names alone are discouraged.

Proposed Schema (Draft)

Building on existing PET Pharmaceuticals field structure:

{
  "Medications": [
    {
      "Name": "Carbamazepine",
      "RxNormCode": "2002",
      "ChEBICode": "CHEBI:3387", 
      "ATCCode": "N03AF01",
      "Dose": 200,
      "DoseUnits": "mg",
      "DoseFrequency": "twice daily",
      "Route": "oral",
      "StartDate": "2024-01-01",
      "EndDate": null,
      "Indication": "epilepsy",
      "Notes": "Stable dose for 6 months prior to recording"
    }
  ]
}

Placement in BIDS

Options to discuss:

  1. participants.tsv columns + detailed participants.json definitions - for chronic medications
  2. Session-level sidecar - for medications at time of scan (see BEP026 attn @bep026; Formalize the concept of [{leading entities}_]{entity_plural}.{tsv,json} file(s) #2283 for subject+sessions.tsv proposal),
  3. Dedicated sub-<label>_medications.tsv file per subject/session - for complex regimens
  4. Extension/Migration from existing Pharmaceuticals field from PET to other modalities

Questions for Community Discussion

  1. Which vocabularies should be REQUIRED vs RECOMMENDED?

    • Should we mandate at least one coded identifier, or allow free-text fallback?

  2. How to handle temporal aspects?

    • Chronic medications vs. acute administration during scan
    • Historical medication exposure (relevant for pharmacological neuroimaging)

  3. Privacy considerations?

    • Medication lists can be identifying - guidance on de-identification?

  4. Compatibility with existing PET Pharmaceuticals fields?

    • Extend or replace current structure?

  5. Animal studies?

    • Same schema for veterinary drugs? Different vocabularies needed?

References

Call for Contributors

We are looking for contributors with expertise in:

  • Pharmacology/clinical pharmacy
  • Clinical data standards (HL7, CDISC)
  • Neuroimaging (particularly clinical iEEG, PET pharmacological studies)
  • Animal research (veterinary pharmacology)

Please comment below if interested in contributing to this BEP.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions