Skill UI

This skill deploys the OpenMed Model Context Protocol (MCP) server, exposing critical clinical NLP tools such as NER, PII extraction, and de-identification. It enables AI coding agents and chat clients to interact with OpenMed's capabilities via simple tool calls, running locally on-device to ensure user data privacy and security. Supports both stdio and Streamable HTTP deployment modes.

This skill performs the crucial ETL process of mapping OpenMed-extracted, terminology-coded clinical facts (conditions, drugs, measurements) into the OMOP Common Data Model (CDM) v5.4 tables. It is used downstream after entity extraction and source coding, transforming raw NLP findings into standardized, OMOP-conformant records for OHDSI analytics.

This skill evaluates de-identification and clinical NER models against rigorous leakage-first release gates (G1a-G8). Instead of relying on aggregate F1 scores, it focuses on detecting residual Protected Health Information (PHI) leakage and enforcing per-label recall floors, which is crucial for HIPAA compliance. Determines if a model is RELEASABLE or QUARANTINED.

This skill facilitates the extraction of population-scale (cohort-level) clinical text from Electronic Health Records (EHR) or FHIR data warehouses using the FHIR Bulk Data Access ($export) operation. It manages the asynchronous job kickoff, polling, and downloading of NDJSON files. The resulting data stream is designed to be fed in batches into OpenMed for mandatory de-identification and Named Entity Recognition (NER) analysis.

Converts raw OpenMed NER outputs (text spans) into structured, standards-compliant FHIR R4 resources (e.g., Condition, Observation, MedicationStatement). It provides mechanical helpers for building CodeableConcepts using major coding systems like SNOMED, LOINC, and ICD-10, ensuring high interoperability with EHR systems and registries.

This utility separates clinical entity extraction from clinical coding. It processes de-identified text, utilizing OpenMed to identify key clinical spans. These spans are then programmatically mapped into structured FHIR R4 resources (like Condition, Observation, and MedicationStatement) and assembled into a Bundle. This process ensures deterministic, standardized data output suitable for interoperability and medical record structuring.

Utilizes OpenMed's analyze_text to perform sophisticated Named-Entity Recognition (NER) on clinical and biomedical text. This tool extracts structured entities such as diseases, drugs, anatomy, genes, and proteins. It provides precise character spans and model confidence scores, making it essential for building robust, privacy-first clinical Natural Language Processing (NLP) pipelines.

This skill extracts critical patient metadata (PHI) from DICOM headers using pydicom, and recursively pulls structured report narrative text from DICOM-SR content trees. It is essential for ingesting medical images into NLP pipelines, enabling subsequent de-identification and clinical text analysis.

Use openmed.extract_pii to detect and retrieve specific types of Protected Health Information (PHI) or Personally Identifiable Information (PII) spans (names, dates, MRNs, etc.) from clinical text. Unlike deidentification, this function returns the original text and the detected spans, allowing users to audit, preview redactions, normalize labels, or route spans to a custom anonymizer without altering the source data. It is ideal for pre-processing and policy enforcement.

This skill extracts crucial Social Determinants of Health (SDOH) — such as housing instability, food insecurity, and transportation barriers — from unstructured clinical notes. It maps the identified text spans to standardized ICD-10-CM Z-codes (Z55–Z65). It is essential for health-equity analytics, building risk stratification tools, and identifying coding gaps by converting narrative text into queryable, discrete data points.

This skill facilitates the ingestion of unstructured clinical notes and reports from FHIR R4 EHR APIs (e.g., Epic, Cerner). It handles complex tasks including paginating large search results, decoding base64 attachments (PDF, text), and extracting narrative content. The extracted text is then passed to the OpenMed pipeline for HIPAA-compliant de-identification and subsequent NLP analysis, enabling structured data insights from raw clinical records.

Synthea generates statistically realistic, fully synthetic patient records in multiple standards (FHIR R4, C-CDA, CSV). This tool is essential for developers and researchers needing safe, shareable test data for OpenMed pipelines, CI fixtures, or specialized de-identification leakage gate test sets. It ensures zero risk of exposing real Protected Health Information (PHI).