TRACER Model Predicts Heart Failure Hospitalization from Low-Resolution Data

The TRACER model utilizes a Transformer with Contrastive Event Representation to predict hospitalizations in irregularly sampled telemonitoring data. It employs time-aware embeddings and contrastive pre-training to identify anomalies in biomarker sequences from heart failure patients.