Data Structure

JSON Reference

When you request the json output format (set output_config.format to json), the Convert and Bulk Convert endpoints return the FinalOutput schema. It is a flat, source-cited structure that captures patient demographics and one or more encounters, with the clinical entities detected inside each encounter.

For the standards-based alternative, request the fhir format instead and see Clinical Resources. The JSON schema carries additional enrichment that FHIR does not, including per-field source citations (page numbers and bounding boxes), MEAT evidence, and ICD-10 and HCC coding detail.

Top-level structure

Field	Type	Description
`version`	string	nHance extraction schema and product version.
`patient_demographics`	array of PatientDemographics	One entry per page on which patient demographics were detected.
`encounters`	array of Encounter	Encounters extracted and enriched from the document.

Patient demographics

Field	Type	Description
`first_name`	string	Patient first name.
`middle_name`	string	Patient middle name.
`last_name`	string	Patient last name.
`date_of_birth`	string	Date of birth.
`age`	string	Age as written in the document.
`gender`	string	Patient gender.
`medical_record_number`	string	Medical record number (MRN).
`address`	string	Patient address.
`phone_number`	string	Contact phone number.
`email`	string	Contact email.
`insurance_provider`	string	Insurance provider name.
`insurance_id`	string	Insurance member ID.
`primary_language`	string	Primary language.
`page_number`	integer	Page the demographics were found on. Nullable.
`bounding_boxes`	object	Per-field bounding boxes, keyed by demographic field name.

Encounter

Field	Type	Description
`encounter_id`	integer	Unique identifier for the encounter.
`start_page`	integer	First page of the encounter in the document.
`end_page`	integer	Last page of the encounter in the document.
`note_type`	string	Type of clinical record. See allowed values below.
`dos`	array of DateWithBoundingBoxes	Dates of service, each paired with bounding boxes.
`admission_date`	DateWithBoundingBoxes	Admission date paired with bounding boxes.
`discharge_date`	DateWithBoundingBoxes	Discharge date paired with bounding boxes.
`telehealth_encounter`	boolean	How the visit was conducted. `true` for a telehealth (virtual) visit; `false` for an in-person, face to face visit.
`visit_type`	string	Type of visit, for example in-person (face to face), telehealth, outpatient, or inpatient.
`providers`	array of Provider	Providers named in the encounter.
`signed_provider`	array of SignedProvider	Providers who signed the note.
`facility_name`	string	Facility name.
`specialty`	string	Provider or encounter specialty.
`reason_for_visit`	string	Reason for the visit.
`vitals`	array of Vitals	Vital-sign readings.
`lab_results`	array of LabTest	Laboratory results.
`diet`	string	Diet notes.
`cognitive_status`	string	Cognitive-status notes.
`diagnoses`	array of Diagnosis	Diagnoses with ICD-10 codes and MEAT evidence.
`combination_codes`	array of CombinationCode	ICD-10 combination codes derived from diagnosis pairs.
`medications`	array of Medication	Medications.
`procedures`	array of Procedure	Procedures.
`social_history`	SocialHistory	Social history.
`family_history`	array of FamilyHistoryItem	Family history.
`allergies`	array of Allergy	Allergies.
`care_plan`	CarePlan	Care plan.
`symptoms`	array of Symptom	Symptoms.
`diagnostic_tests`	array of DiagnosticTest	Diagnostic tests.
`valid_encounter`	boolean	Whether this is a billable, real encounter.

note_type allowed values: progress note, labs, facesheet, discharge summary, emergency record, diagnostic imaging, therapy note, unknown, others.

Vitals

Field	Type	Description
`blood_pressure`	string	Blood pressure.
`heart_rate`	string	Heart rate.
`respiratory_rate`	string	Respiratory rate.
`temperature`	string	Temperature.
`height`	string	Height.
`weight`	string	Weight.
`bmi`	string	Body mass index.
`oxygen_saturation`	string	Oxygen saturation.
`page_number`	integer	Page the vitals were found on.

Diagnoses

Field	Type	Description
`name`	string	Diagnosis name.
`description`	string	Full description or context sentence from the document.
`section_header`	string	Section header the diagnosis was found under.
`page_number`	integer	Page the diagnosis was found on.
`code1`	string	Primary ICD-10 code. Nullable.
`code2`	string	Secondary or alternative ICD-10 code. Nullable.
`meat`	MeatLinkages	MEAT evidence and linked entities.
`combination_codes`	array of string	Combination ICD-10 codes this diagnosis participates in.
`bounding_boxes`	array of BoundingBox	Where the diagnosis name appears on the page.

MEAT linkages

MEAT (Monitor, Evaluate, Assess, Treat) evidence for a diagnosis.

Field	Type	Description
`context`	string	Free-text MEAT evidence from the document.
`medication`	array of string	Medications linked to this diagnosis.
`lab`	array of string	Lab tests linked to this diagnosis.
`procedure`	array of string	Procedures linked to this diagnosis.
`diagnostic_test`	array of string	Diagnostic tests linked to this diagnosis.

Combination codes

Field	Type	Description
`code`	string	The combination ICD-10 code.
`main_diagnosis`	string	Name of the main diagnosis in the pair.
`main_code`	string	ICD-10 code of the main diagnosis.
`associated_diagnosis`	string	Name of the associated diagnosis in the pair.
`associated_code`	string	ICD-10 code of the associated diagnosis.

Medications

Field	Type	Description
`name`	string	Medication name.
`section_header`	string	Section header the medication was found under.
`medication_class`	string	Drug class, for example ACE inhibitor.
`dosage`	string	Dosage.
`frequency`	string	Frequency.
`route`	string	Route of administration.
`prescribed_date`	string	Date prescribed.
`prescriber`	string	Prescriber.
`status`	string	Status, for example active.
`page_number`	integer	Page the medication was found on.

Lab results

Field	Type	Description
`name`	string	Lab test name.
`result`	string	Result value.
`reference_range`	string	Reference range.
`status`	string	Result flag, for example high.
`collection_date`	string	Collection date.
`page_number`	integer	Page the lab result was found on.

Procedures

Field	Type	Description
`name`	string	Procedure name.
`section_header`	string	Section header it was found under.
`year_of_procedure`	string	Year of the procedure.
`page_number`	integer	Page the procedure was found on.

Diagnostic tests

Field	Type	Description
`name`	string	Diagnostic test name.
`result`	string	Result.
`section_header`	string	Section header it was found under.
`date_performed`	string	Date performed.
`page_number`	integer	Page the diagnostic test was found on.

Allergies

Field	Type	Description
`name`	string	Allergen name.
`reaction`	string	Reaction.
`severity`	string	Severity.
`page_number`	integer	Page the allergy was found on.

Care plan

Field	Type	Description
`recommendation`	string	Recommendation.
`treatment`	string	Treatment.
`therapy`	string	Therapy.
`follow_up`	string	Follow-up instructions.
`page_number`	integer	Page the care plan was found on.

Field	Type	Description
`occupation`	string	Occupation.
`education`	string	Education.
`marital_status`	string	Marital status.
`housing`	string	Housing situation.
`abuse`	array of AbuseItem	Substance-use entries.
`page_number`	integer	Page the social history was found on.

Each abuse entry has type, status, and frequency string fields.

Family history

Field	Type	Description
`relation`	string	Relation to the patient.
`condition`	string	Condition.
`age_of_onset`	string	Age of onset.
`page_number`	integer	Page the entry was found on.

Symptoms

Field	Type	Description
`name`	string	Symptom name.
`description`	string	Description.
`section_header`	string	Section header it was found under.
`page_number`	integer	Page the symptom was found on.

Providers

providers entries have name and page_number, plus bounding_boxes. signed_provider entries add credentials (for example MD) and signed_date.

Source citations

Most objects carry a page_number and bounding_boxes so every value can be traced back to its exact place on the page.

A BoundingBox is an array of four numbers, [x1, y1, x2, y2]. Values may be normalised floats between 0 and 1, or pixel integers. On patient demographics, bounding_boxes is an object keyed by field name; on diagnoses and providers it is an array of boxes.

Example

A trimmed json result for a single-encounter progress note.

{
  "version": "3.1.0",
  "patient_demographics": [
    {
      "first_name": "R",
      "last_name": "Matthews",
      "date_of_birth": "1978-03-04",
      "gender": "female",
      "medical_record_number": "RM-2284",
      "page_number": 3,
      "bounding_boxes": { "last_name": [[0.09, 0.14, 0.24, 0.17]] }
    }
  ],
  "encounters": [
    {
      "encounter_id": 1,
      "start_page": 3,
      "end_page": 3,
      "note_type": "progress note",
      "specialty": "Cardiology",
      "providers": [{ "name": "A. Okonkwo, MD", "page_number": 3 }],
      "vitals": [{ "blood_pressure": "148/92", "heart_rate": "102", "oxygen_saturation": "94%" }],
      "diagnoses": [
        {
          "name": "Heart failure, unspecified",
          "code1": "I50.9",
          "code2": null,
          "meat": { "context": "Cardiology follow-up in 2 weeks", "medication": ["Lisinopril"] },
          "page_number": 4
        }
      ],
      "medications": [
        { "name": "Lisinopril", "dosage": "20 mg", "frequency": "daily", "route": "PO", "status": "active" }
      ],
      "allergies": [{ "name": "Penicillin", "reaction": "Urticaria", "severity": "high" }],
      "valid_encounter": true
    }
  ]
}