Nudge Care

FOR DEVELOPERS · APIS

Structured

medical data

from any PDF

Structured

medical data

from any PDF

One documents API. Send any medical PDF — lab results, diagnostic report, or clinical note — and we route it through the right extraction pipeline.

Book a demo

See the review app

Lab

Diagnostic

LAB.PDF

P.1/2

HCA Healthcare UK

PATHOLOGY REPORT · 2025-09-22

BIOCHEMISTRY (BLOOD)

Total Protein

60–80 g/L

Albumin

35–50 g/L

Globulin

20–40 g/L

Total Bilirubin

<21 µmol/L

ALT

<33 U/L

AST

<32 U/L

Non-HDL

Cholesterol

2.9

— mmol/L

unstructured · multi-column · scanned

POST

10-30s

response.json

200 OK

// extracted & mapped

{

"document_id": "doc_8a3f",

"type": "lab_report",

"observations": [

{

"name": "Total Bilirubin",

"value": 30,

"unit": "umol/L",

"range": "<21",

"flag": "high",

"loinc": "1975-2",

"confidence": 0.97

{

"name": "Non-HDL Cholesterol",

"value": 2.9,

"unit": "mmol/L",

"loinc": "43396-1",

"confidence": 0.94

}

// + 12 more

]

}

Lab

Diagnostic

LAB.PDF

P.1/2

HCA Healthcare UK

PATHOLOGY REPORT · 2025-09-22

BIOCHEMISTRY (BLOOD)

Total Protein

60–80 g/L

Albumin

35–50 g/L

Globulin

20–40 g/L

Total Bilirubin

<21 µmol/L

ALT

<33 U/L

AST

<32 U/L

Non-HDL

Cholesterol

2.9

— mmol/L

unstructured · multi-column · scanned

POST

10-30s

response.json

200 OK

// extracted & mapped

{

"document_id": "doc_8a3f",

"type": "lab_report",

"observations": [

{

"name": "Total Bilirubin",

"value": 30,

"unit": "umol/L",

"range": "<21",

"flag": "high",

"loinc": "1975-2",

"confidence": 0.97

{

"name": "Non-HDL Cholesterol",

"value": 2.9,

"unit": "mmol/L",

"loinc": "43396-1",

"confidence": 0.94

}

// + 12 more

]

}

ONE ENDPOINT

Upload clinical files,

get structured results

Upload clinical files,

get structured results

We classify the document type and route it through the right extraction pipeline automatically. Through the API on a folder, and see the magic happening.

One endpoint, multiple pipelines

Post your clinical files and we'll do the rest. Every response includes a type field so your service knows which extraction shape to expect.

→

Lab results route to LOINC mapping with values, units, ranges

→

Diagnostic reports route to SNOMED CT + ICD-10/11 with exam classification

→

Clinical notes route to condition, medication, and recommendation extraction

PDF

HCA Lab.pdf

PDF

MRI Wrist.pdf

PDF

Consult.pdf

PDF

Discharge.pdf

CLASSIFY

& ROUTE

Lab pipeline

LOINC · values · ranges

Diagnostic pipeline

SNOMED · ICD-10 · exam

Notes pipeline

conditions · meds · plan

TWO WAYS TO CONSUME

Validate the output on your system or on our App

Same engine, same document types. Choose how the structured output reaches your service — programmatically, or after a clinician has signed it off

API

Direct, programmatic

POST a document, get structured JSON back. You handle the review step on your side; your team, your UI, your workflow.

Review queue

Reviewed by a clinician

The output lands in our review queue app for clinical sign-off. Bulk-download validated documents, or pull them as a reviewed API stream.

CLINICAL VALIDATION

The review app

When we send you an API key, we also send you an invite to our playground. Drop in a real PDF, see the extraction, switch between table view and raw JSON, decide if it's good enough before you write a single line of integration.

Lab results

Diagnostic report

Nudge APIs

Lab results

Clinical documents

EXTRACTED FILES

▼ HCA Healthcare UK La...

17 items

Date: 2025-09-22

RECENT EXTRACTS

Mar 11, 01:26 PM

{}

Extraction Results

HCA HEALTHCARE UK 2025.PDF

VALIDATED

Table

JSON

BIOMARKER

VALUE

UNIT

RANGE

Frequently Asked Questions

No. The APIs are extraction tools. They convert unstructured PDFs into structured data, with confidence scores. They don't make clinical recommendations and they're not approved as a medical device. Output should be reviewed by a clinically-aware person before being used in care — through our playground, your own team, or our clinical operations team.

OINC for lab values. SNOMED CT and ICD-10/11 for diagnostic findings, conditions, and clinical notes. We can map to additional ontologies on request — talk to us.

Two ways. Either bulk-download all validated documents from the review queue app as JSON or CSV — useful for batch processing — or pull them via our reviewed-stream API, which returns only documents that have been signed off by a clinician. Each reviewed document includes the audit trail: who reviewed it, what changed during review, and when.

By default no — we process and discard. If you'd rather keep an audit-ready record you can manage and purge yourself, we offer that too. The review queue requires temporary retention until sign-off, after which the same delete-or-keep options apply. We don't share data with third parties and we don't train models on customer data.

High enough that it saves you months per project versus building it yourself. Try it on a sample of your real documents in the playground and judge it on your data — that's why the trial exists. We will publish benchmarks as we evolve our algorithm.

JSON today. FHIR R4 on the roadmap — our internal data model is already FHIR-native, so the API layer is the remaining work. Talk to us if FHIR is a hard requirement; we can prioritise.

It depends of the type of the document, if it's lab results or clinical notes. We prioritize accuracy over performance.

Yes. On Production, you can add the review queue UI for your own team to use — same playground you tried in trial, with bulk export and a reviewed-stream API on top. Or, if you'd rather not staff review at all, our clinical team in Greece handles it for you on Enterprise.

Still have questions?We're here to help.