HealthVerity Notes

HealthVerity Notes is a dataset of de-identified clinical EHR narratives designed for use in real-world data (RWD) research, drug development, and health economics and outcomes research (HEOR) studies. It is intended for researchers and data teams who need access to the unstructured, free-text content of clinical records — such as clinician observations, symptom descriptions, and disease progression notes — that structured coded data does not capture.

The dataset is delivered as a modular, standalone product anchored to HealthVerity IDs (HVIDs), making it linkable with other data sources in the HealthVerity ecosystem including claims, EHR, and lab data. De-identification is applied using detect-and-replace logic while preserving the grammatical and linguistic structure of the original documentation.

Core Dataset Characteristics

• Delivered as a standalone, modular dataset that requires no additional technical implementation.
• Each record is anchored with an HVID, enabling joins with claims, EHR, and lab datasets within the HealthVerity ecosystem.
• De-identification uses detect-and-replace logic that retains sentence structure and linguistic integrity.
• Preserves clinically relevant detail including symptoms, disease trajectory, and provider impressions.
• Grammatical structure is maintained to support more accurate natural language processing (NLP), including entity recognition and predictive modeling.

Clinical Content and Research Applications

• Captures the clinical reasoning behind care decisions, including undocumented symptoms and progression timelines that do not appear in structured fields.
• Surfaces context that coded data misses, such as clinician observations and qualitative assessments of patient status.
• Supports use cases in precision medicine, real-world evidence generation, and patient journey analysis.
• Designed to enrich existing RWD workflows by adding a narrative layer to structured data assets.

NLP Integration and Processing

• HealthVerity partners with NLP vendors to provide condition-specific processing of the notes data.
• Supported NLP applications include adverse event detection, severity classification, and symptom clustering.
• The preserved sentence structure of de-identified notes is intended to improve the accuracy of NLP outputs.
• Unstructured free-text content can be converted into structured signals suitable for downstream analysis and modeling.

HealthVerity Notes is interoperable with the broader HealthVerity Marketplace, where users can build custom patient cohorts and combine multiple data sources. The dataset is positioned as a complement to structured data assets, adding clinical narrative depth to RWD strategies across therapeutic research and outcomes studies.