Pull structured medical facts, entities, relationships, citations, into your own tools. BETA
A dedicated API surface exposing the platform's structured evidence graph: entities, relationships, and provenance, so your own legal or claims tools can consume cited medical facts directly. Every node returned traces back to the source page or DICOM slice it came from, built on the same graph behind Medical Record Knowledge Graph, not a separate dataset.
One graph, exposed as an API.
Query by case, entity type, or date range and get back the same structured evidence graph the platform builds internally: diagnoses, treatments, providers, and imaging, along with the relationships between them and the exact page or DICOM slice each one traces to.
Not the same thing as the Medical Data API, or the Knowledge Graph.
The Medical Data API gives you the platform's documents and extracted fields. Medical Record Knowledge Graph is that structure, browsable inside the product. The Evidence Graph API is the same graph's entities, relationships, and provenance, exposed as a dedicated API surface for your own tools to consume directly.
Extracted fact and inferred relationship are never returned as the same thing.
The API exposes extracted, cited data, facts directly stated in the record and traceable to a page or DICOM slice, alongside relationships the model has inferred between them. The schema tags every node and edge with which one it is.
Your tools should never present an inferred relationship to a claims decision or a court filing as if it were an established fact. The API gives you the distinction; what to do with an inferred edge stays a human decision on your side of the integration.
From case to queryable graph.
Three steps, no separate export pipeline to maintain.
Call the API with a case ID, entity type, or date range.
Entities, relationships, and provenance, structured and ready to consume.
Syncs incrementally as new documents are processed on the case.
Who builds on it.
Teams with their own systems that need cited medical facts, not just a PDF export.
Feed cited facts into case-management and drafting tools automatically.
For law firmsSync structured evidence straight into the claims system of record.
For carriersKeep the evidence graph in sync across client-specific workflows.
For TPAsPull cited findings into scheduling and reporting systems already in place.
For IME orgsEvidence Graph API, answered.
It's a dedicated API surface that exposes the platform's structured evidence graph, the entities, relationships, and provenance extracted from a case, so your own legal or claims tools can consume cited medical facts directly instead of re-reading the record.
The Medical Data API surfaces the platform's documents, pages, and extracted fields. The Evidence Graph API sits on top of that layer and exposes the structured graph built from it, the entities and their relationships, for tools that need connections between facts, not just the raw fields.
The Medical Record Knowledge Graph is the in-product view of that same graph inside Medrecords AI. The Evidence Graph API exposes that graph's entities, relationships, and provenance as a dedicated API surface, so it can be consumed by systems outside the platform.
Yes. Every node and edge is typed in the schema as either an extracted fact, something directly stated in the record and cited to its source page or DICOM slice, or an inferred relationship the model has proposed. Inferred edges are never returned as if they were established facts.
Yes, in beta. The Evidence Graph API is live and testable now against real cases, and we're refining the schema and endpoints hands-on with early integration partners. If your use case is a good fit, we'll work with you directly.
Query the evidence graph on one of your own cases.
Join the beta and try the API against a real file. Handled under our BAA; never used to train a model.