Extract data from clinical notes
Unstract parses clinical notes, from typed progress notes to faxed patient charts, and returns each diagnosis, vital sign and medication the clinician documented, with the line it came from, to your EHR or clinical data warehouse.
pt to bring glucose log next visit - PN
PN-SOAP (REV 2024/03) · PAGE 1 OF 2
{ "patient": "Halvorsen, Margaret", "date_of_birth": "03/09/1961", "date_of_service": "10/28/2026", "provider": "Priya Nandakumar, MD", "blood_pressure": "142/88", "diagnosis": "Type 2 diabetes with hyperglycemia", "icd10": "E11.65", "new_medication": "Empagliflozin 10 mg daily", "follow_up": "3 months, repeat A1C", "confidence": 0.93}
Trusted by healthcare and revenue cycle teams around the globe
The chart you need is rarely the one your system printed
Notes from your own clinicians export cleanly. Notes from outside practices, specialists and older charts reach the HIM office in whatever shape the sender kept them.
A specialist's progress note faxed twice before it reached you
A handwritten plan with one dose struck and rewritten
An EHR printout, a dictated note and a paper SOAP template
A whole chart scanned as one PDF, every note type inside
A vitals flowsheet stapled behind the visit notes
Clinical documentation specialists and nurse reviewers read every page of a chart to find one diagnosis or one medication change, and a review that waits on that reading delays the authorization or the quality report behind it.
Unstract reads each clinical note, pulls the diagnoses, vitals, medications and follow-up you name, links every value to its line in the note, and sends the record to your EHR or clinical data warehouse.
Read the note, structure what the clinician wrote, route the chart
LLMWhisperer reads the note, Document Studio returns the structured record, and Flow carries it into prior authorization or a review queue. Each product runs alone, so a chart review team can begin with the parsing on its own.
Read clinical notes with every SOAP heading kept in place
A progress note is written under its own headings: subjective, objective, assessment, plan. Clinical note parsing only helps if a blood pressure stays under Objective and a family member's diagnosis never lands in the patient's own assessment. LLMWhisperer keeps the layout of the page, so the vitals row reads as a row and the plan reads as a list. It reads typed notes, dictated transcripts, handwritten addenda and charts that went through a fax machine more than once.
Pass the text to your own LLM, or make it the first step of a full chart review run.
Learn more about LLMWhispererTurn each clinical note into one record of what was documented
Describe the record once: patient, date of service, provider, chief complaint, vitals, each diagnosis as written with any ICD-10 code printed beside it, medications with dose and change, follow-up and signature. SOAP note data extraction then returns the same JSON from an EHR printout, a dictated note or a paper template. Checks you set travel with the record: a medication in the plan that is missing from the medication list, a date of service that differs from the fax date, an unsigned note.
Teams searching for NLP to extract diagnosis from clinical notes often want the reasoning too. Unstract returns the diagnosis exactly as the clinician wrote it and the line it came from. It does not assign a code the note never printed. Deploy it as an API behind your review tools, or as an ETL pipeline over the folder where scanned charts land.
Learn more about Document StudioRun prior authorization from the order to the payer's decision
Describe how your prior authorization team uses clinical notes, the orders they support and each payer's criteria, in plain language, and Flow's agents build the workflow from that description. You approve it, and every request then runs the same way: the order and its progress notes picked up, the diagnoses and history the payer asks about pulled from the chart, the request assembled and tracked until the payer answers, with approvals posted to your EHR.
Criteria that are not met, documentation that is missing from the chart and urgent cases stop for a prior authorization coordinator or a nurse reviewer, with the reason attached. Unstract finds what the clinician documented. It does not decide medical necessity or make a clinical judgment.
Learn more about FlowParse a chart's worth of notes today. Plan the prior authorization run with us.
Start for free and extract a folder of de-identified progress notes. A run that feeds your EHR or clinical data warehouse starts with a consultation.
Progress notes, SOAP notes and the rest of the patient chart
Documents
- Progress notes
- SOAP notes
- Doctor's notes
- History and physical notes
- Consultation notes
- Care plans
- Behavioral health assessments
- Post-procedure follow-up notes
- Discharge summaries
- Nursing notes
- Patient charts as scanned PDFs
- Any other clinical note in your charts
Fields captured
Processes these notes feed
Built for the teams who read the chart after the visit
Outside charts and scanned records indexed by note type and date of service, with the key fields ready for your EHR or clinical data warehouse.
Each documented diagnosis found with the line that supports it, so a query to the provider starts from the exact sentence in the note.
Medication changes, follow-up dates and care plan goals pulled from every visit note, so outreach lists reflect the latest plan.
Diagnoses, prior treatment and the dates they started gathered from the progress notes before a reviewer checks them against the criteria.
Less page-turning per chart, faster reviews
Clinical notes are treated as the most sensitive record in the chart
A progress note holds the patient's diagnoses, medications and, in a behavioral health assessment, the clinician's own notes on the patient's mental health. That combination reveals conditions a patient may not have told their family or their employer.
- Encrypted in transit and at rest.
- Clinical notes are never used to train models.
- Access by role, note type and department, so behavioral health notes can be limited to the reviewers cleared to see them.
- Processed notes are deleted on the retention schedule your organization sets.
- Self-hosted deployment for health systems that keep every chart inside their own environment.
What HIM, documentation and review teams ask about clinical notes
Look for an API that keeps the layout of each note, reads scans and faxes, and returns every value with its place on the page. Unstract offers that in two parts: LLMWhisperer turns each chart page into layout-preserved text, and Document Studio deploys an endpoint per note type that returns structured JSON. POST a chart and the record comes back with flagged fields marked for review.
Unstract does. You describe the record once, and Document Studio returns the same fields from a progress note, a care plan or a behavioral health assessment, whichever practice wrote it. A chart scanned as one PDF is split by note type first, so each note gets its own record.
Unstract returns the date of service of each note, the start date of each medication where the clinician wrote one, and procedure dates from post-procedure follow-ups. Across a chart, that gives a dated list of treatment that utilization reviewers can check against a payer's criteria.
Document Studio can return a complications field from each daily note, filled with what the surgeon or nurse wrote, such as a wound infection or a fever on day two, along with the sentence it came from. A note that records none returns an empty field, so a missing entry never reads as a complication.
Yes. Family history comes back as its own field, with each relative and condition as the clinician wrote them, so a father's heart attack never appears in the patient's own diagnoses.
Yes. LLMWhisperer reads handwriting alongside printed text, including a plan written in pen with a dose crossed out, and it reads dictated transcripts as ordinary typed notes. A value it cannot read with certainty goes to a reviewer instead of into the record.
No. Unstract extracts what the clinician documented. A diagnosis comes back as written, and an ICD-10 code appears only when it is printed on the note. Coding and clinical decisions stay with your coders and clinicians.
Document Studio deploys as an API your developers call, or as an ETL pipeline that reads a folder of scanned charts and writes records to a data warehouse you already run. Flow runs the full prior authorization process and posts results to your EHR.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant. Notes are encrypted in transit and at rest, never used to train models, and kept only as long as your retention schedule allows. Access can be limited by note type, and health systems can self-host. The controls are published in the Trust Center.
A documentation or HIM lead can start for free and extract fields from a few progress notes the same day. For a prior authorization run, we build it with your team in a consultation, around your note types, your payers and your EHR or clinical data warehouse.
Bring a de-identified chart to the consultation
Progress notes, a care plan, a faxed specialist note and a post-op follow-up. We parse them in the session and show which values would have gone to a reviewer, and why.