Harmonized ECG features improve cross-dataset clinical prediction

Researchers have developed a harmonized and interpretable feature representation for electrocardiogram (ECG) waveforms to improve cross-dataset clinical prediction. This approach, called FeatureDB, aims to reduce performance degradation when models trained on one hospital's data are applied to another's, addressing differences in protocols and measurements. The study evaluated this method on tasks like heart failure classification and mortality prediction, finding that the harmonized features maintained competitive performance and demonstrated greater stability in cross-dataset transfer compared to end-to-end deep learning models like ConvNeXt. AI

RANK_REASON Research paper published on arXiv detailing a new method for harmonizing ECG features. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Alberta Cohort
• Anita Khalafbeigi
• ConvNeXt
• electrocardiography
• FeatureDB
• MIMIC-IV
• XGBoost

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →