Terminology Server

Terminology Server is a medical terminology management product from John Snow Labs designed to convert free-text medical language into standardized codes. It is built for healthcare and life sciences organizations that need accurate, auditable, and scalable terminology mapping across major coding systems including ICD-10-CM, ICD-10-PCS, ICD-9-CM, SNOMED-CT, LOINC, RxNorm, CPT4, MedDRA, HCPCS, NDC, UMLS, MeSH, NCIT, ATC, CVX, HPO, HGNC, and ICDO3.

The platform supports both interactive user-interface workflows and full programmatic access via API, and can be deployed on-premise or through AWS and Azure Marketplace. It is intended for enterprise environments with regulatory and compliance requirements, including HIPAA-compliant configurations.

Text-to-Code Conversion and Search Capabilities

• Accepts free-text medical input and returns corresponding medical codes from supported vocabularies.
• Automatically detects spelling errors and abbreviations in user queries, with an option to accept or ignore suggested corrections.
• Context Resolution Search ranks and prioritizes results based on the user's specific intent, reducing irrelevant results and improving precision for specialized domains.
• Boolean Search Operations allow use of AND/OR logic for exact text matching to refine code results.
• Advanced Filtering supports filtering by domain (e.g., Condition/Device, Drug, Gender), OMOP Standards Concepts Only, Concept Validity, and Confidence Score.
• Term Weight and Scope controls allow users to adjust search precision based on confidence levels and surrounding context, and can be combined with vocabulary filters.

Concept Maps and Cross-Vocabulary Mapping

• Supports locating a specific medical code within curated coding systems and viewing its equivalent in a different vocabulary.
• Supports crosswalks such as SNOMED CT to ICD-10 and LOINC to SNOMED mappings.
• Cross-vocabulary mapping is available on demand within agentic AI pipelines via MCP Server Integration.

Schema-Based Clinical Document Coding

• Predefined entity schema covers diagnoses, procedures, medications, and observations across all major code systems.
• Direct mapping to standard terminologies including ICD-10, SNOMED CT, CPT-4, RxNorm, LOINC, ATC, and MedDRA, with no post-processing required.
• Deterministic output ensures identical inputs always produce identical coded results, supporting auditability.
• Supports high-throughput batch coding of large volumes of clinical documents with sub-second per-document latency.

Document Search

• Document Accurate Search allows users to define or modify entity labels at runtime using a built-in pretrained clinical zero-shot entity recognition model.
• Entities are mapped to different code systems based on user-defined mappings.

Custom and Prebuilt Value Sets

• Users can upload custom value sets via CSV to create tailored medical code lists for specific clinical or operational needs.
• Version control is supported for custom value sets to track changes and maintain consistency.
• Custom value sets can be applied as filters during search to target specific code subsets.
• A prebuilt public library of value sets is available, including CMS Chronic Conditions, Potentially Planned ICD-10-PCS Procedures, and Acute Diagnosis ICD-10-CM.
• Prebuilt value sets are versioned and traceable, supporting reproducibility across time points and audit cycles.
• Value Set constraints can be applied directly within Default Search, Map Codes, or Document Search modes.

MCP Server Integration for Agentic AI Pipelines

• Terminology Server search modules (Default, Map Codes, Document Search) can be registered as MCP-compliant named agent tools without custom API wrappers.
• Enables autonomous AI agents to perform real-time concept normalization, cross-vocabulary mapping, Value Set validation, and code lookup at inference time without human-in-the-loop intervention.
• Supports cross-vocabulary mappings such as SNOMED CT to ICD-10, LOINC to SNOMED, and RxNorm to NDC on demand.
• A fully documented Jupyter notebook demonstrating end-to-end agentic terminology workflows is included.

Enterprise API Access and Security

• Full API access covers all functionality available through the user interface, enabling custom workflow integration.
• Standard vocabularies receive regular updates to maintain accuracy and compliance with the latest code releases.
• Security features include encryption in transit and at rest, role-based access control for administrative functions, and comprehensive audit logging.
• Compatible with single sign-on via SAML/OAuth when integrated through the UI.
• No patient identifiers or clinical notes are stored; only standard terminologies and user-defined mappings are retained.
• Queries remain within the organization's secure network when deployed on-premise or in a private cloud, with no data transmitted to John Snow Labs or third parties.

Deployment and Scalability

• Available for on-premise installation and through AWS Marketplace and Azure Marketplace.
• Architecture supports horizontal scaling to handle high concurrency and has been tested with millions of concept entries.
• Optimized for sub-second query times in typical deployments, including real-time code lookups for simultaneous EHR transactions.
• Embedding-based search is accelerated with vector indexing techniques to maintain speed as new terms and embeddings are added.
• Batch operations leverage distributed processing where available.
• Licensing provides full access to all product features with no limitation on the number of users or searches.

Onboarding resources include step-by-step installation documentation and comprehensive API code samples. Support is available via email, phone, and Slack, with paying customers receiving a private Slack channel. Academic and non-profit pricing discounts are available upon request.