{}

Healthcare Billing Automation: Extract Data From EOBs & Medical Bills

Explanation of Benefits (EOBs)Medical billsSuperbillsItemized statements

The problem

Billing teams manually re-key claim numbers, billed/allowed/paid amounts, and procedure codes from EOBs and provider bills into billing or patient-accounting systems. Reconciling payer adjustments by hand is slow, and it gets harder to keep accurate at volume.

How Dokyumi solves it

Define a schema for exactly the EOB/medical-bill fields you need — claim number, provider, service date, billed/allowed/paid amounts, patient responsibility, procedure codes — and get a dedicated endpoint. Submit scanned or PDF EOBs and get back validated JSON.

Because you define the schema, you capture only the fields you actually need — a simple way to minimize how much sensitive data you collect in the first place.

For handling of protected health information, see our security page — Dokyumi uses a HIPAA-aware architecture, and a BAA is available on the Enterprise plan.

Example schema and result

A real schema shape for healthcare billing documents, and the JSON you get back from one extraction.

Schema — GET /api/v1/schemas

{
  "slug": "eob-parser",
  "name": "EOB Parser",
  "description": "Extracts claim and billing data from Explanation of Benefits statements",
  "ocr_mode": "standard",
  "fields": [
    {
      "key": "claim_number",
      "type": "string",
      "required": true
    },
    {
      "key": "provider_name",
      "type": "string",
      "required": true
    },
    {
      "key": "service_date",
      "type": "date",
      "required": true
    },
    {
      "key": "billed_amount",
      "type": "currency",
      "required": true
    },
    {
      "key": "allowed_amount",
      "type": "currency",
      "required": false
    },
    {
      "key": "paid_amount",
      "type": "currency",
      "required": false
    },
    {
      "key": "patient_responsibility",
      "type": "currency",
      "required": false
    },
    {
      "key": "cpt_codes",
      "type": "array",
      "required": false
    }
  ]
}

Result — POST /api/v1/extract

{
  "id": "ext_hb3q7m",
  "status": "completed",
  "schema": "eob-parser",
  "data": {
    "claim_number": "CLM-2026-771234",
    "provider_name": "Lakeside Family Medicine",
    "service_date": "2026-05-08",
    "billed_amount": 480,
    "allowed_amount": 310,
    "paid_amount": 248,
    "patient_responsibility": 62,
    "cpt_codes": [
      "99214",
      "90471"
    ]
  },
  "confidence": {
    "claim_number": 0.99,
    "billed_amount": 0.98,
    "cpt_codes": 0.93
  },
  "meta": {
    "processing_time_ms": 1340,
    "page_count": 2,
    "ocr_cached": false
  }
}

How it works

1

Define the fields you need

Claim number, amounts, procedure codes — only what your billing workflow actually uses.

2

Get a dedicated endpoint

Scoped to that schema alone.

3

Submit EOBs and medical bills

PDF or scanned image, one API call.

4

Reconcile automatically

Structured JSON with confidence scores flows into your billing or patient-accounting system.

Healthcare Billing FAQ

Is Dokyumi HIPAA-compliant?+
Dokyumi is built with a HIPAA-aware architecture: encryption in transit and at rest, access controls, and schemas designed to help you minimize what you collect. A Business Associate Agreement (BAA) is available on the Enterprise plan — contact hello@dokyumi.com before sending PHI. See /security for details.
Can I limit which patient fields get captured?+
Yes — you design the schema, so you only extract the fields your workflow needs. Leaving a field out of the schema means Dokyumi never captures it.
Does Dokyumi integrate with my EHR or billing system?+
Not with a native pre-built connector today. Extractions are delivered as JSON via the API or a webhook, so you integrate it into whatever system you already use.
Can it read handwritten superbills?+
OCR accuracy on handwriting is limited compared to typed or printed documents; typed EOBs and itemized statements extract most reliably.
What happens with a low-confidence field, like a hard-to-read procedure code?+
The extraction still completes, but the response flags low-confidence fields so you can route them for manual review instead of trusting them blindly.

Related use cases

Insurance

Pull named insured, policy numbers, coverage limits, and claim details from dec pages, ACORD forms, and claims paperwork.

Accounts Payable

Extract vendor, line-item, and total data from invoices and receipts — no template building, no AWS Textract setup.

Tax Prep

Turn client-submitted W-2s, 1099s, and 1040s into structured data — or hand clients a branded upload portal for intake season.

Stop re-keying explanation of benefits (eobs) by hand.

100 free extractions every month — no credit card, no AWS account, no setup hell.