Item — HN Reader

Worldwide medical standards implementations with Rust for FHIR, openEHR, SNOMED, etc. Official specifications work well with Claude Fable, test automation, and storage using relational databases.

Fast Healthcare Interoperability Resources (FHIR) using Rust:

https://github.com/fhir-rust/fhir-rust

Open Electronic Patient Record (openEHR) using Rust:

https://github.com/openehr-rust/openehr-rust

Systematised Nomenclature of Medicine (SNOMED) using Rust:

https://github.com/snomed-rust/snomed-rust

I have work in progress creating stores for PostgreSQL, SQLite, MySQL, MariaDB, MSSQL, Oracle.

1 comments

Do any of those specs address de-identification or pseudonymisation, or is that always left to the implementer? I've hit exactly that gap with an app I am working on so would appreciate to hear what you have to say on the matter.

Yes in some senses, but no from a compliance perspective. This is AFAIK and IANAL and YMMV etc.

For example: The standard FHIR $export operation can support extensions that take de-identification config details as additional parameters. The standard openEHR can query then return non-id fields. The standard SNOMED codes can be rolled up e.g. high-specificity codes can be aggregated into a low-specificity code.

But: for compliance/legal/auditing reasons, all organizations that I know of personally do not rely exclusively on these protocols for de-identification. Instead, the organizations place their de-indentification processes in a kind of "clean room" that does many more kinds of cleansing, redacting, aggregating, chaffing, filtering, and the like.