Extracting data from attending physician statements
Insurers and state disability programs ask the treating physician for an attending physician statement, each on its own form, completed by hand or in a fillable PDF. Life underwriters and disability claims examiners order them by the thousand. They arrive by fax or through provider portals.
Examiners need the diagnosis and its ICD-10 codes, the date care began, the treatment frequency and the date disability began. The claims system uses them to set the benefit start date and the next review, and an underwriter uses them to rate the medical risk.
What makes attending physician statements hard
Many dates with distinct meanings
One statement carries the first visit, the most recent visit, the date disability began and an expected return to work, each beside its own caption.
Written by the physician
Physicians often complete the statement by hand, writing the diagnosis and the care dates into boxes sized for a few characters.
Answers are ticked boxes
Treatment frequency, hospital stays and whether the condition is work related are answered with ticked boxes, several options to a line.
Parse, extract, verify
Parse
Parse turns the document into reading-order Markdown and a hierarchy of blocks (text, tables, figures), and grounds each block to its exact location on the page.
Extract
You define the fields in a JSON schema, and Extract returns their values in a consistent, structured format, each linked back to where it appears in the Markdown.
Verify
Each value links back to where it appears, down to the character range in the Markdown, and the page and bounding box in the source document, so every result can be verified.
One attending physician statement, end to end
About this document: The physician's certificate from the California Employment Development Department's sample disability insurance claim form, filled with specimen data. Three portrait pages, with the practitioner and the care timeline on page 1. Page 2 holds the diagnosis codes and findings, and page 3 the history and the signed certification. Page 1 is featured. It asks for the same core facts as an insurer's statement, and the patient is a specimen.
Processing objective: The objective is to establish who certified the claim and the timeline of care. The practitioner's license type and issuing state identify the certifier. The care dates and treatment interval set the benefit period, and each value grounds to its own row of the form.
Demonstration fields: License type, License issuing state, Under care from, Treatment interval, Date disability began
The document
3 of 3 pages
Page 1shown below
Page 2
Page 3
Where each value came from
practitioner.license_type
{
"value": "MD",
"page": 1,
"range": {
"end": 1676,
"start": 1629
},
"box": {
"xmin": 0.05991,
"ymin": 0.29503,
"xmax": 0.49405,
"ymax": 0.33387
}
}practitioner.license_issuing_state
{
"value": "CA",
"page": 1,
"range": {
"end": 1627,
"start": 1555
},
"box": {
"xmin": 0.40586,
"ymin": 0.25775,
"xmax": 0.94655,
"ymax": 0.28065
}
}treatment_period.under_care_from
Printed on the page as From 01/25/2025 To M M D D Y Y Y Y, normalized on extraction.
{
"value": "2025-01-25",
"page": 1,
"range": {
"end": 2244,
"start": 2208
},
"box": {
"xmin": 0.06436,
"ymin": 0.63357,
"xmax": 0.5018,
"ymax": 0.65744
}
}treatment_period.treatment_interval
{
"value": "Monthly",
"page": 1,
"range": {
"end": 2361,
"start": 2349
},
"box": {
"xmin": 0.06623,
"ymin": 0.66384,
"xmax": 0.93401,
"ymax": 0.6851
}
}disability.date_disability_began
Printed on the page as [x] Yes - enter date disability began 01252025, normalized on extraction.
{
"value": "2025-01-25",
"page": 1,
"range": {
"end": 2625,
"start": 2579
},
"box": {
"xmin": 0.06273,
"ymin": 0.724,
"xmax": 0.57855,
"ymax": 0.74539
}
}How to try it yourself
Everything on this page is reproducible. The sample, its schema, and the Parse and Extract API outputs are published. You can repeat the same two calls from whichever surface you already build on.
- Claude Skilldescribe the job; the agent writes it
- CLIno code at all
- MCPexpose ADE docs as an agent tool
- REST APIany language
- Pythonlandingai-ade
- TypeScriptlandingai-ade
Documents into data
Accurate
The best accuracy on the difficult documents other tools get wrong, such as messy scans, complicated tables, checkboxes and handwriting.
Auditable
Every answer cites its sources. Values come back with an audit trail (page, bounding box, character range) so a human can review them.
Agentic
Applies multi-step reasoning, and plans before it acts. Handles documents it has never seen before and adapts to variations.
API-first
Composable APIs that drop into any tech stack and can be called from anywhere. Easy to build on and embed in your pipelines and products.
Affordable
Cost-competitive rates with complete price transparency. Spend controls before you run, and every job reports its exact cost.
Enterprise-grade
Built for production and scale from day one. Fully observable and monitored, running securely in our cloud or yours.
Built for high-volume, high-throughput production
- Zero data retention
- Priority rate limits
- Enhanced support
- VPC and on-prem deployment
Available on Team and Enterprise plans. VPC and on-prem deployment require an Enterprise plan.
