Interview summary

Thomas Beale is one of the original architects of openEHR, having worked with Sam Heard and others from the mid-1990s through to the early 2000s to develop what became the archetype model at the heart of the standard. His interview is the most technically foundational of the set - he explains not just what openEHR does but why the problem it solves is structurally different from almost any other domain in software engineering.

The core insight, which took years to arrive at, is that healthcare data is orders of magnitude more complex than data in any other sector, and that complexity is not reducible by better engineering - it requires a different approach entirely. Traditional database engineering, where you design a schema and add tables and columns for every data type, simply cannot scale to a domain with potentially hundreds of thousands of distinct data items that are also constantly evolving. The answer Thomas and Sam arrived at in 1997 was the separation of the reference model from the archetype layer: a small, stable database schema that knows about the concept of observations and orders, combined with a formal modelling layer - archetypes - that clinical specialists can build and maintain without touching the database. This allows the cognitive model of a nurse or doctor to be represented computably, which is something no other publicly known methodology achieves.

The community this has generated - over 3,000 registered members across 100 countries, with a core active group of 500 to 600 - is, for Thomas, proof that the concept works. Clinical specialists contribute in small fragments of time, building and reviewing models that a hundred contributors from 20 countries might have shaped over months. The blood pressure archetype alone contains 25 data points, each precisely defined, named, coded, and unitised - a level of clinical fidelity that no IT-led process could produce.

The barriers to realising the full vision are not technical. A shared medication list visible to every clinician who ever sees a patient is not a hard problem to solve, Thomas observes - and yet it does not exist after 40 years of discussion. The reason is the shape of the industry: incumbent vendors with petabytes of data in proprietary formats, healthcare institutions too risk-averse to change running systems, and procurement processes that default to buying another big product rather than building platform infrastructure. His conclusion is that the change required is primarily educational - governments, ministries of health, and procurement bodies need to understand that a different model is possible, and that standardised APIs and open platforms could support an entire ecosystem of specialist applications rather than a single vendor trying to do everything. The destination is a single patient-centred record, in a persistent open format, that follows the person through every encounter for life. That is the dream, and the gap between here and there is political and commercial, not technical.

Back to video library