OpenMed v2.5.0¶
OpenMed 2.5 expands the Python SDK with clinical privacy and extraction previews, local privacy and audit controls, bounded multimodal intake, FHIR and OMOP validation, and metadata-only training orchestration. The release baseline is v2.3.0; there is no intervening v2.4.0 release.
This page describes the candidate. Registry publication, immutable package tags, and deployed documentation are verified separately after publication.
Clinical processing and interoperability¶
- Opt-in clinical-preserving privacy policies with explicit language, category, role, and review controls; German address and clinical-phrase regressions; and full-document ONNX batching with bounded cancellation.
- Source-aligned context, medication quantities, measurements, relations, temporal events, and timelines. Ambiguous or incomplete quantities are withheld. These are preview APIs requiring task and language qualification.
- A canonical grounded-span FHIR export facade, conservative DiagnosticReport export, base R4 and US Core checks, and validators for reference targets, profile declarations, and Observation extensions.
- OMOP CDM 5.4 measurement, procedure, visit, observation-period, and NOTE_NLP exporters; cohort integrity checks; and FHIR R5 round-trip fidelity reports.
- Local ValueSet expansion, explicit caller-configured terminology endpoint delegation, and a provenance-aware terminology cache. Restricted vocabulary content and the official HL7 Java validator remain external to the package.
- A Persian PII language pack and structural locale-tag normalization.
Privacy and audit controls¶
The release adds local policy schemas, composition and migration checks, minimum-necessary field selection, field-level FHIR/OMOP policies, tabular risk and schema-drift reports, surrogate integrity checks, and structured redaction and idempotence evidence.
Audit tooling covers access scope and expiry, key custody and rotation, evidence integrity, replay, lineage and freshness, report size and cardinality, waiver lifecycle, exception budgets, policy coverage and simulation, and release-evidence aggregation. Reports use declared identifiers, counts, offsets, hashes, and bounded metadata rather than protected source values.
Encrypted surrogate mappings, local deletion planning and verification, retention cleanup, pre-push and CI privacy scanning, dataset-upload guards, and session-end trace scrubbing are explicit local operations. The optional Snowpark adapter supports caller-managed warehouse processing; it does not turn the core package into a network service.
Multimodal, agent, and runtime additions¶
- Bounded BMP, GIF, WebP, and WAV metadata inspection; RF64 refusal coverage; batch summaries, cache keys, and pre-decode asset limits.
- One ordered accept-or-abstain preflight report that composes manifest, media, profile, limit, and digest checks. Missing metadata never counts as successful validation.
- Memory-streamed Tesseract OCR and verified raster PDF/PNG exports with source-line and redaction-fidelity checks.
- Strict provider-result and run-summary parsers, opaque correlation and artifact references, and typed governance identifiers.
- A local privacy-proxy app with an injected transport, per-request placeholder restoration, and streaming placeholder buffering.
- ONNX label checks before optional runtime imports, actual tensor batching, execution limits, and independent model-batch admission.
- Federated scheduling, round status, update metadata and aggregate metrics; PHI-free retraining slice ranking and recipe proposals; and local compute, cost, energy, and carbon budget evidence.
Compatibility and installation¶
Read the 2.3-to-2.5 migration guide, including registry compatibility and the versioned model-state migration. Python 3.10+ remains supported. The core dependency set stays unchanged; Snowpark is optional and the grounding-validate extra does not bundle the Java validator.
After publication, the intended coordinates are:
pip install --upgrade "openmed==2.5.0"
pip install --upgrade "openmed[hf,fhir]==2.5.0"
npm install openmed@2.5.0
Container consumers select ghcr.io/maziyarpanahi/openmed:v2.5.0. The Helm application version and default tag are 2.5.0; explicit image selection uses image.tag=v2.5.0.
The new clinical, privacy-governance, and orchestration APIs described above are Python capabilities. Updating the npm, Apple, and Android package versions does not imply these new APIs run natively on every platform.
Qualification boundary¶
Synthetic regression tests and package checks establish the behavior they exercise. They do not establish clinical efficacy, deployment-specific identifier recall, regulatory compliance, or clinical-device suitability. Clinical outputs require qualified review and must not automatically determine diagnosis, treatment, or billing.
This SDK candidate preserves the existing model pointer targets and model artifacts. The registry state representation changes independently of those targets. A model promotion requires its own staged golden and public SHIELD benchmark evidence, signed gates, and authorization under the release-stream policy.
The complete change record is in the changelog.