Towards a shared language for health interventions
WHO · International Classification of Health Interventions (ICHI)
The WHO Family of International Classifications has long pursued a coherent, interoperable framework for describing human health. CodAs is a small experimental contribution to that larger effort.
Open CodAs →A vision decades in the making
The WHO Family of International Classifications (WHO-FIC) encompasses three complementary systems: the International Classification of Diseases (ICD), which describes what people suffer from; the International Classification of Functioning, Disability and Health (ICF), which describes how they function; and the International Classification of Health Interventions (ICHI), which describes what is done for them.
Together, these three classifications form the conceptual backbone of a globally interoperable health information system – a vision that has been developed, debated and refined by hundreds of experts, institutions and national health authorities over many years, under the auspices of the WHO and its collaborating centres worldwide.
ICHI: designed for interoperability
Among the three WHO-FIC classifications, ICHI occupies a distinctive position. Its post-coordination architecture – the systematic combination of independent axis elements (Action, Target, Means) and extension codes – was designed to express a vast range of health interventions from a finite set of building blocks.
This structural openness makes ICHI inherently linkable: to other WHO-FIC classifications, to national procedure catalogues, to nursing interventions, and to classifications beyond the WHO-FIC family entirely. The coding logic is not lookup-based but compositional – closer in nature to a grammar than to an index.
A timely intersection
The global rollout of ICHI coincides with a period of rapid development in Large Language Models (LLMs). These systems have demonstrated a notable capacity to work across natural languages, formal rule systems, and structured vocabularies simultaneously – capabilities that are directly relevant to the challenge of health intervention coding.
This raises a question worth investigating: to what extent can LLMs support the application of a compositional classification like ICHI, and under what conditions? That question is not yet answered. But it seems worth asking – and exploring systematically.
What CodAs does – and does not do
CodAs is an experimental coding assistant. It attempts to interpret a free-text description of a health intervention and to construct a plausible ICHI post-coordinated code from the available axis elements, following the coding rules defined in the ICHI Reference Guide.
Because LLMs operate across languages without requiring translation, CodAs accepts requests in any language and in any level of technical detail. A description in plain everyday language is as valid as one in clinical terminology. This is one of the properties that may prove most relevant to a global rollout of ICHI – reducing the linguistic and professional barriers to engagement with the classification.
⚠️ An experimental tool – use with care
CodAs is at an early stage of development. It can produce plausible and useful coding suggestions, but it can also produce errors. No warranty is given for the accuracy of any generated code.
All results should be reviewed by a qualified person familiar with ICHI. CodAs is not a substitute for professional coding expertise, and it is not approved for use in clinical documentation without independent verification. The ICHI verification tool allows stem codes to be checked against the official WHO dataset.
Please note that ICHI is an evolving international standard, continuously developed by the WHO. CodAs updates its underlying dataset twice yearly; the most recent update was applied on March 13, 2026. Coding suggestions may not yet reflect the latest WHO revisions.
Try CodAs
CodAs is freely available. It requires an Anthropic API key, which is billed directly to the user's own account. Describe a health intervention in your own words and language – and see what CodAs makes of it.