Applying natural language processing to electronic health record data—From text to triage

Applying natural language processing to electronic health record data—From text to triage
JAMA Network Open; Grace K. Sun, BS; Andrew P. Ambrosy, MD; 11/24
Most information about a patient’s clinical status, disease progression, and response to treatment lies in qualitative clinician documentation in the electronic health record (EHR). The New York Heart Association (NYHA) classification was developed to standardize functional status assessments and treatment decisions ... [but] ... due to inconsistent implementation in routine care, much of the critical information remains in unstructured EHR data that is difficult to capture and analyze. Natural language processing (NLP) is an emerging tool that uses artificial intelligence to process unstructured or semistructured free-text data, such as the embedded assessments of HF symptom status in clinician documentation. NLP, a field of artificial intelligence that focuses on understanding, interpreting, and generating human language, is capable of evaluating these data and providing large-scale insights into patient progress and treatment response, with some limitations.