Machine learning-based identification and validation of aging-related genes in cardiomyocytes from patients with atrial fibrillation.

Background: Aging is a key risk factor for atrial fibrillation (AF), a prevalent cardiac disorder among the elderly. This study aims to elucidate the genetic underpinnings of AF in the context of aging.

Methods: We analyzed 12,403 genes from the GSE2240 database and 279 age-related genes from the CellAge database. Machine learning algorithms, including support vector machines and random forests, were employed to identify genes significantly associated with AF.

Results: Among the genes studied, 76 were found to be potential candidates in the development of AF. Notably, four genes - PTTG1, AR, RAD21, and YAP1 - stood out with a Receiver Operating Characteristic Area Under the Curve (ROC AUC) of 0.9, signifying high predictive power. Logistic regression, validated through 10-fold cross-validation and Bootstrap resampling, was determined as the most suitable model for internal validation.

Conclusions: The discovery of these four genes could improve diagnostic accuracy for AF in the aged population. Additionally, our drug prediction model indicates that bisphenol A and cisplatin, among other substances, could be promising in treating age-associated AF, offering potential pathways for clinical intervention.