EMR Models Predict Atrial Fibrillation Risk
Researchers in Taiwan develop and externally validate machine-learning models to predict atrial fibrillation during hospitalization among ischemic stroke patients, using electronic medical records from Landseed International Hospital (3,988 patients) and Chia-Yi Christian Hospital (5,821 patients) in a 2026 JMIR Med Inform study. The study compares nine algorithms, integrates TF-IDF text features, and finds ensemble models outperform others, with E/A ratio, left atrial size, and age as top predictors, supporting targeted AF screening.
Key Points
- 1Built ensemble machine-learning models predicting atrial fibrillation using EMRs from two hospitals (3,988 and 5,821 patients)
- 2Integrated structured variables with TF-IDF text features, improving performance in one dataset and generalizability
- 3Highlight common top predictors: E/A velocity ratio, left atrial size, and age for targeted AF screening
Scoring Rationale
Multi-institutional validated ML study provides solid evidence, but incremental novelty and limited geographic scope moderate impact.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems