Skip to content

Clinical Data Transformer

Transform your unstructured clinical notes into standardized, coded medical data

Get API Access View Documentation

Overview

The Clinical Data Transformer (CDT) is an API that processes free-text clinical documentation and extracts structured medical information. Using advanced AI, CDT identifies medical concepts within clinical notes and maps them to standardized SNOMED CT terminology, and reads the medications a note lists with their dosing, coding them to the ATC classification. Both are returned together, enabling integration with healthcare systems and data analytics pipelines.

Automated Extraction

Extract medical concepts from clinical notes automatically using state-of-the-art language models.

SNOMED CT & ATC Coding

Map extracted concepts to standardized SNOMED CT terminology (International Edition 2026-01-01) and medications to the WHO ATC classification (2026 Release).

FHIR R4B Output

Receive results as FHIR R4B Bundles for seamless healthcare system interoperability.

Temporal Reasoning

Associate clinical findings with dates and track disease progression over time.

Clinical Concepts

CDT analyzes clinical text and identifies structured medical concepts across multiple categories. Medications are read in the same pass and returned beside the concepts; see Medications below.

Included Concepts

Category Description Examples
Disorders All diagnoses, diseases, and conditions - current, historical, confirmed, differential, or resolved Aortic valve stenosis, Type 2 diabetes, history of TIA, resolved hepatitis B
Procedures Diagnostic and therapeutic procedures, with what became of each: carried out, under way, awaited, or cancelled PCI, appendectomy, CT scan, chemotherapy, physical therapy
Symptoms Patient-reported symptoms and chief complaints Lumbar pain, dysphagia, muscle weakness, vertigo
Findings Clinical observations including normal and abnormal results Lymphadenopathy, intact reflexes, CTAB, proteinuria
Morphologic Abnormalities Anatomical or structural abnormalities Pleural effusion, hepatomegaly, stenosis, sclerosis
Allergies All documented allergies and hypersensitivities Penicillin allergy, latex hypersensitivity, shellfish allergy
Negated concepts Concepts the note names in order to deny them, returned with a refuted status rather than dropped "no evidence of infection", "ruled out PE", "hepatitis excluded"

Excluded Concepts

Exclusion Reason Examples
Numeric values Quantitative measurements without qualitative interpretation "creatinine 1.2 mg/dL", "LVEF 40%", "BP 120/80"
Family history Conditions in relatives only "father with colon cancer", "family history of CAD"
Bare recommendations A procedure merely floated rather than arranged. A procedure that is scheduled, arranged or awaited is returned, with a preparation status "recommend MRI", "consider surgery"
Administrative terms Non-clinical documentation "follow-up appointment", "referral to specialist"

SNOMED CT Reference

A complete list of the SNOMED CT concepts that can be extracted by CDT is available in the CDT FHIR Implementation Guide.

Extracted Data Fields

For each identified concept, CDT returns:

  • SNOMED CT Code - Standardized clinical terminology code from the SNOMED CT International Edition (2026-01-01)
  • SNOMED CT Term - Human-readable term from the SNOMED CT vocabulary
  • Status - For a disorder, finding or allergy, how certain the note is: confirmed, provisional, differential, or refuted when the note denies the concept. For a procedure, what became of the act: completed, preparation, in-progress or not-done
  • Date / End Date - Temporal information when available
  • Text Position - Character offsets where the concept was found

Medications

Every medication a note mentions is returned alongside the concepts, in the order in which it appears. Medications are read as prescriptions rather than as concepts: what matters is the product, where the patient stands with it and how it is taken.

Extracted Medication Fields

For each medication mentioned, CDT returns:

  • ATC Code - Classification code and its display from the 2026 WHO ATC classification, at the level the classification reaches the product
  • Name and Strength - The drug as the note names it, brand name included, with the strength as printed
  • Form and Route - How the medication is dispensed and how it is taken, e.g. tablet taken orally
  • Status - Where the patient stands with it: active, stopped, on hold, completed or planned
  • Dosing - How often it is taken and how much at each moment, whether it is taken only when needed, and the dates or duration the prescription runs for
  • Text Position - Character offsets where the mention was found

Uncoded Medications

A medication the ATC classification does not reach is still returned, without a code. What the note lists is never dropped for want of a code.

Languages

The CDT API has been tested and validated against clinical notes in English language.

Other Languages

While other languages may work with the API, they have not been formally tested. Additionally, SNOMED CT coding is currently limited to the International Edition only — country-specific extensions are not yet supported.

Coming Soon

Support for additional language extensions is planned for future releases.

Ready to Get Started?

Request access to the CDT API beta program and start transforming clinical notes today.

Request API Access Already have an invite code? Sign up

Contact

For questions, support, or partnership inquiries: