TraCeR: Transformer-Based Competing Risk Analysis with Longitudinal Covariates

This article introduces TraCeR, a transformer-based model for competing risk analysis. The use of transformers suggests an attempt to capture complex temporal dependencies in longitudinal data. The application to competing risk analysis is significant, as it addresses scenarios where multiple events can occur, and the occurrence of one event can preclude others. The paper's focus on longitudinal covariates indicates an effort to incorporate time-varying factors that influence the risk of events.