CAE Product Release Notes

Release Date: October 2026

About

Context-Aware Extraction (CAE) is a family of domain-specific APIs that analyze full unstructured clinical notes and return clinically appropriate, coding-relevant structured outputs. Unlike sentence-level extraction approaches that evaluate each sentence independently, CAE leverages document-level understanding across multiple sections and sentences to resolve clinical problems and procedures to their most specific coded representations.

This approach reduces manual review, improves coding specificity, and delivers higher-fidelity structured data for downstream coding, revenue cycle, CDI, analytics, and AI workflows.

What's New in Our Latest Release

• Introduced the CAE Procedure endpoint.
• Added document-level procedure extraction and normalization.
• Added support for CPT, HCPCS, ICD-10-PCS, and SNOMED CT mappings.
• Added procedure-specific context handling, including laterality, body site, approach, and technique.
• Expanded explainability support across procedure outputs.

Supported Domains

Problem Domain

• Document-level problem extraction and normalization
• ICD-10-CM focused coding support
• Evidence-based explainability

Procedure Domain

• Document-level procedure extraction and normalization
• Procedure status identification (performed, historical, ordered)
• Support for laterality, body site, approach, and technique when documented
• Mapping to IMO Lexicals, CPT®, HCPCS, ICD-10-PCS, and SNOMED CT®
• Evidence-based explainability

Model Evaluation Results

CAE v2.0.0 was evaluated using a 100-note gold standard dataset designed to represent real-world clinical documentation. Performance is reported using Precision, Recall, and F1 Score, where F1 balances both precision and recall.

Release Notes