Extract data from itemized bills and superbills
Unstract captures every line of a hospital itemized bill, superbill or patient invoice, scanned, faxed or sent as a PDF, and passes the checked charges to your billing or claims system.
| DATE REV SERVICE | UNITS | CHARGE |
|---|---|---|
| DETAIL OF CHARGES | ||
| 10/27 0450 ER visit, level 4 · 99284 | 1 | 1,365.00 |
| 10/27 0301 Comp metabolic panel · 80053 | 1 | 148.00 |
| 10/27 0305 CBC w/ auto diff · 85025 | 1 | 62.00 |
| 10/27 0352 CT abd/pelvis w/ contrast · 74177 | 1 | 2,940.00 |
| 10/27 0305 CBC w/ auto diff · 85025 | 1 | 62.00 |
| 10/27 0120 Room & board, med/surg | 2 | 4,180.00 |
| 10/28 0636 Ondansetron 4 mg inj · J2405 | 2 | 38.00 |
| 10/28 0258 IV saline 1,000 mL · J7030 | 2 | 96.00 |
| ACCOUNT SUMMARY | ||
| Total charges | 8,891.00 | |
| Insurance payment | (5,412.60) | |
| Contractual adjustment | (2,206.40) | |
| Patient balance due | 1,272.00 | |
CBC twice on 10/27? ask PFS - RN
Questions about this statement: 530-555-0126 · Page 1 of 2
{ "facility": "Larkspur Valley Medical Center", "patient": "Vance, Teodora", "account_number": "88104273", "admit_date": "10/27/2026", "discharge_date": "10/29/2026", "total_charges": 8891, "insurance_paid": 5412.6, "patient_balance": 1272, "duplicate_cpt": "85025", "confidence": 0.95}
Trusted by healthcare and revenue cycle teams around the globe
No two providers print a bill the same way
A hospital's billing system prints one layout, a physician practice another, and a patient sends whatever the front desk handed them.
Faxed by the hospital, scanned again by the plan
A superbill with codes circled and one diagnosis crossed out
Three providers, three places to look for the charges
The claim, the itemized bill, the EOB and a receipt in one file
A five-day stay that prints hundreds of supply and drug lines
Bill reviewers retype charge lines into a spreadsheet to find the duplicates and the lines missing from the claim, so a long hospital bill sits in the queue while its payment deadline gets closer.
Unstract reads each bill, returns every date, code, unit and charge as one record, flags the lines that repeat or don't match the claim, and sends the result to your billing or claims system.
Read the bill, capture each charge, route the exceptions
LLMWhisperer reads the bill, Document Studio turns it into a checked record, and Flow runs medical bill processing around both. Each product works on its own, so a review team can start with the extraction and add the full run later.
Read itemized bills with every line, code and amount on its own row
A hospital itemized bill packs a date of service, a revenue code, a description, a CPT or HCPCS code, units and a charge onto one printed line, and repeats that for pages. LLMWhisperer keeps the columns lined up, so a unit count never slides into the charge column and a bracketed payment keeps its minus sign. Faxed copies, skewed scans and phone photos of patient invoices come out as the same clean text.
Send the text to your own LLM, or make it the first step of a bill review run.
Learn more about LLMWhispererTurn each itemized bill into one structured list of charges
Describe the record once: facility, patient, account number, admission and discharge dates, and for every line the date of service, revenue code, CPT or HCPCS code, units and charge. A hospital statement, a superbill from a specialist and a patient invoice from a clinic all come back in that same JSON. Checks run as the lines come out: the lines must add up to total charges, payments and adjustments must leave the printed balance, and a line that repeats on the same date arrives flagged with the line it copies.
Deploy it as an API that your bill review tool calls, or as an ETL pipeline over the folder where faxed bills land.
Learn more about Document StudioRun itemized bill review from the fax line to the reviewer's queue
Describe how your team reviews itemized bills and superbills in plain language, and Flow's agents build the workflow from that description. You approve it, and every bill then runs the same way: picked up with the claim it supports, read line by line, compared with the claim and the contract rates you load, and sent on to your billing or claims system when nothing is out of place.
A duplicate line, a charge missing from the claim or a total that doesn't foot stops for a bill reviewer, with the lines side by side. Unstract reads the charges. It does not assign codes, price services or judge medical necessity; those calls stay with your reviewers and your payment policies.
Learn more about FlowCapture the lines on today's bills. Plan the review run with us.
Start for free and extract a batch of itemized bills yourself. A run that checks bills against claims and feeds your billing or claims system starts with a consultation.
Hospital statements, superbills and the invoices patients send in
Documents
- Hospital itemized bills
- Itemized statements of charges
- Patient superbills
- Patient invoices
- Patient statements
- Physician and professional fee bills
- Emergency department bills
- Ambulance and transport bills
- Laboratory and imaging bills
- Pharmacy and medical supply receipts
- Explanations of benefits sent with a bill
- Any other bill format a provider prints
Fields captured
For the teams that read charges line by line
Itemized statements ready the moment a payer or a patient asks for one, with every line matching what was billed on the claim.
Superbills from many client practices captured into one format, so charge entry no longer waits on someone reading circled codes.
Long hospital bills compared with the claim before payment, with repeated and unbundled lines already marked for the reviewer.
Bills from out-of-network providers and employee reimbursement requests read the same way, ready for the plan's payment rules.
Bills reviewed sooner, charges keyed less
Charges and health details, protected as PHI
An itemized bill pairs a patient's name and account number with every test, scan and drug they received, and the price charged for each. Read together, the lines describe the diagnosis without ever naming it.
- Encrypted in transit and at rest.
- Bills and their charge lines are never used to train models.
- Access by role, client and payer, with every corrected line recorded against the bill it came from.
- Processed bills are deleted on the retention schedule your compliance team sets.
- Self-hosted deployment for health plans and billing companies that keep PHI inside their own environment.
What billing and bill review teams ask first
A claim form puts a fixed number of fields in fixed boxes. An itemized bill is a table that runs for pages in whatever layout the provider's billing system prints. Itemized bill OCR has to keep every row intact, so each charge stays with its date, revenue code and units. LLMWhisperer reads it that way and Document Studio returns each row as its own line in the record.
Yes. LLMWhisperer shows which boxes on a superbill are ticked or circled and which are left blank, so superbill data extraction returns only the CPT and ICD-10 codes the provider selected, with units, fees and the date of service.
Yes. Every page is read and every line is returned, page breaks and repeated headers included. Document Studio then checks that the lines add up to the total charges printed on the last page.
Each line is compared with every other line on the bill. Two lines with the same date of service, code and units come back flagged, with both line numbers, so the reviewer can confirm a true repeat or a second test that was actually ordered.
No. Unstract reads the charges as the provider printed them and checks them against the claim and the rates you supply. Assigning codes, pricing services and deciding medical necessity stay with your reviewers, coders and payment policies.
Yes. A patient's photo of a clinic invoice or a pharmacy receipt goes through the same extraction as a hospital statement, and comes back in the same record with the provider, dates of service and amounts paid.
Yes. Document Studio deploys an endpoint per document type: POST an itemized bill and get the header and every charge line back as JSON, with flagged lines marked. The same extraction also runs as an ETL pipeline over a folder or an S3 bucket.
Any system with an API or a file import. You shape the JSON to match your billing or claims system, and Flow connects to more than 100 systems when you run the whole review there.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant. Bills are encrypted in transit and at rest, never used to train models, and deleted on your retention schedule. Health plans and billing companies can self-host. The Trust Center lists the safeguards.
For extraction, a batch of real bills and a list of the fields your reviewers use. Start for free and upload them the same day. For the full review run, we build it with your team in a consultation around your claims, contract rates and billing or claims system.
Bring your longest itemized bills to the consultation
Hospital statements, superbills and the patient invoices that arrive by fax or phone photo. We read them in the session and point out every line a reviewer would have stopped on.