What it does
- Classifies an incoming document (lab report, prescription, discharge summary, insurance policy, invoice and more) and flags handwritten and non-medical inputs
- Extracts structured fields per document type: tests, results, units and reference ranges from lab reports; diagnoses, medications, dosages and advice from prescriptions; policy and claim details from insurance documents
- Codes the output to standard ontologies: LOINC and UCUM for lab tests, SNOMED CT for clinical concepts, our drug database for medications
- Redacts personally identifiable information in text and in the document image itself
- Returns results as JSON and as an HL7 FHIR document bundle
Read more
- Releasing Parrotlet Open Models: Open Weights for Medical Document Intelligence – Parrotlet-v 2.5 Pro, Med Doc Classifier and Document PII Redactor
- Parrotlet-V Lite (4B) – our first open vision LLM for medical records
- Lab-Ready and Prescription-Perfect: Eka Care’s Small LLMs vs. Industry Giants
- Extracting Structured Information from Lab Reports: Challenges and Learnings
Models and tools
- Parrotlet-v 2.5 Pro · Parrotlet-V-Lite-4B · Med Doc Classifier · Document PII Redactor
- Evaluation set: Medical Records Parsing Validation Set
- Reference: LOINC-linkable lab tests
Try it out
- Live demos: Medical records · Classifier · PII redactor
- API: Medical Document Parsing
- Entity linking: Medical Entity Codification