Incident atrial fibrillation and its risk prediction in patients developing COVID-19: A machine learning based...

Area of Science:

• Gerontology
• Cardiology
• Infectious Diseases

Background:

• Elderly patients with multiple comorbidities are at increased risk for adverse outcomes from COVID-19.
• Incident atrial fibrillation (AF) is associated with worse outcomes in the general elderly population.
• Identifying COVID-19 patients at high risk for incident AF is crucial.

Purpose of the Study:

• To investigate incident AF risks in elderly patients with and without COVID-19.
• To analyze the impact of cardiovascular and non-cardiovascular comorbidities on AF development.
• To compare main effect modeling with a machine-learning approach for predicting incident AF.

Main Methods:

• A prospective cohort of 280,592 elderly US patients was studied over 8 months.
• Incident AF outcomes were examined in relation to COVID-19 status and multimorbidities.
• Both main effect modeling and a machine-learning (ML) approach were employed.

Main Results:

• Multimorbidity, including cognitive impairment, anemia, diabetes, and vascular disease, was associated with COVID-19 onset.
• COVID-19 showed the highest association with incident AF (OR 3.12), followed by heart failure and coronary artery disease.
• The ML algorithm demonstrated superior discriminatory validity and clinical utility compared to main effect models.

Conclusions:

• COVID-19 significantly impacts incident AF development in elderly patients with comorbidities.
• The developed ML approach effectively predicts new-onset AF in COVID-19 patients by considering dynamic multimorbidity changes.