Document classification
Identify the likely document type without replacing the retained original.
OpenMed · document intelligence layer
AI should point to the evidence—not pretend to be the answer.
OpenMed is the source-linked document engine inside the MedicalRecords.in architecture. It creates private machine drafts; it is not an AI doctor, diagnosis service, or autonomous clinical record.
OpenMed processing and account uploads are not active in this public-tools deployment.
Evidence pipeline · architecture previewNot active here
Original source
Retained patient-owned document
OpenMed draft
Entities · dates · sensitive spans
Doctor release gate
Source checked · decision recorded
The useful non-diagnostic work
Reduce document friction without manufacturing clinical certainty.
Structured extraction
Locate dates, providers, clinical entities, and sensitive spans as machine-draft evidence.
Source-span linking
Keep extracted items traceable to the document and source position that produced them.
Packet assembly
Support chronology, duplicate handling, and source-linked visit-pack preparation.
Three states that must never blur together.
ImplementedCode, tests, and synthetic evidence exist at named revisions.
Deployment-validatedThe exact control has been observed in a named deployed environment. Many OpenMed operating controls remain open.
Externally reviewedQualified independent clinical, privacy, legal, and security review is still required for the applicable scope.
The release rule
The original remains the authority. The draft remains private.
Patient-facing clinical content is withheld pending verified-doctor review. Reviewers must be able to inspect the source, resolve blockers, record a decision, and leave an auditable note. That design is implemented in source; staffing, clinical quality, response operations, and deployment evidence remain separate gates.