A new study demonstrates the effectiveness of rule-based natural language processing (NLP) in identifying worsening heart failure events in hospitalized patients, revealing significant variations by age, sex, race and ethnicity, and heart function.
Kaiser Permanente Bernard J. Tyson School of Medicine (KPSOM) faculty members Andrew P. Ambrosy, MD, Associate Professor of Health Systems Science, Alan S. Go, MD, Professor of Health Systems Science, and Kristi Reynolds, PhD, MPH, FAHA, Professor of Health Systems Science, coauthored the study, “Rule-based natural language processing to examine variation in worsening heart failure hospitalizations by age, sex, race and ethnicity, and left ventricular ejection fraction,” published by American Heart Journal. The study analyzes over 44,000 adults diagnosed with heart failure between 2014 and 2019.
NLP methods detected more hospitalizations (12.7 per 100 person-years) compared to regular discharge diagnosis codes (9.3 per 100 person-years). The highest hospitalization rates were seen in adults aged 75 and older, men, and non-Hispanic Black and Hispanic individuals. The study also found that NLP helped better identify heart failure events among Asian/Pacific Islander adults and those with mid-range or preserved heart function, highlighting its potential to improve care for different groups.
