Open-source clinical history framework

A life of medical history, read in seconds.

OpenClinicalHistory turns a patient's scattered, unstructured records into a single anatomical map — every event extracted, placed on the body, and coded to SNOMED CT.

See how it works View on GitHub

A free framework for health organisations to implement and extend — a foundation, not a finished product.

X-ray style anatomical figure with multiple systems highlighted at once — brain, digestive organs, shoulder and knee joints and bladder — each an event resolved to its body region

Why this exists

It started with Melissa.

Melissa has read this page and agreed to her story being told here, by name.

For years, every new specialist meant the same ritual: my wife's history, recited from memory at the start of the appointment. Three open-heart surgeries. A heart valve. A stroke. Frontal-lobe seizures. A growth at the base of her skull. Then breast cancer.

Every doctor needed the whole picture — and every time, the whole picture depended on how well I could tell it.

I first tried to build something for this eight years ago, and couldn't. The hard part — reading a messy discharge letter and actually understanding it — simply didn't exist yet.

Now it does. OpenClinicalHistory is what I couldn't build then: a way for a clinician to see everything that has happened to a patient in seconds, instead of hearing it read aloud.

What it is

A framework, not a product.

OpenClinicalHistory is a reference implementation you host yourself. It hands a health organisation the whole pipeline — vocabulary, ingestion, extraction, classification and the anatomical view — as a foundation to implement, govern and build on. It is deliberately not a patient management system, and it makes no clinical decisions on its own.

What it is

  • A free, open-source framework
  • A complete SNOMED CT classification pipeline
  • An anatomical way to read a whole history at a glance
  • Yours to self-host, adapt and extend

What it isn't

  • A patient management system
  • A hosted service that holds your data
  • A medical device
  • A decision-maker — every event is human-reviewed

How it works

Five stages, from raw records to a coded map.

01

Ingest all of SNOMED CT

The complete SNOMED CT release is imported into a structured database — concepts, descriptions and relationships — giving the system the full international clinical vocabulary to work against.

350,000+ concepts · findings · procedures · body structures
02

Build the anatomical layer database

The SNOMED body-structure hierarchy is mapped to visual body regions, so that any anatomical concept can resolve to a precise place on the figure.

123037004 |Body structure| → body layer
03

Ingest raw, unstructured records

Discharge summaries, operation notes and specialist letters go in as they are. No templates, no forms to fill — the messy source documents themselves.

PDFs · letters · op notes → text
04

Extract the events

Each document is read and broken into discrete clinical events — a diagnosis, a procedure, a finding — then held for human review before anything is committed to the record.

document → events · reviewed, not assumed
05

Classify every event to SNOMED

Each event is bound to a SNOMED concept, and its anatomy resolved through the finding-site and procedure-site relationships — lighting up the exact region on the map.

Condition → 363698007 |Finding site| → body region
The same pass that draws the picture emits a fully coded history — structured, portable and standards-based.

The part I didn't plan

Not a viewer. A classification engine.

I set out to help a doctor understand Melissa faster. What came out the other end was a complete SNOMED classification tool.

The map is what people see first. But underneath it, every event is now bound to SNOMED CT — the international clinical standard — with its anatomy resolved.

That means the output isn't really a picture at all. It's standards-grade coded clinical data: the kind any FHIR-based system can read, exchange and build on.

The visualisation is the interface. The coded history is the product.

Open source

Take it further than I can.

This is a working proof of concept, released free and open. The concept is proven; the hardening — clinical validation, conformance, scale — is exactly where a community and health organisations can carry it further than one person building on evenings.

If you build

The whole pipeline is on GitHub — vocabulary import, ingestion, extraction, classification and the anatomical view. Fork it, break it, improve it.

If you deliver care

It's yours to implement, govern and extend inside your own environment, under your own compliance — not a service you hand your data to.

View on GitHub Get in touch