Line of work

Natural Language Processing for Clinical Phenotyping

This line of work develops rule-based and transformer-based NLP models to extract critical clinical data from unstructured notes. Applications include the extraction of cancer staging, PD-L1 expression, and VTE events.

Anchor result

The VTE-BERT model achieved a precision of 95% and recall of 91% in external validation for detecting incident VTE events.

Related publications

2026
2026
2026
2025
2024
2024
2023
2023