Temporal Validation of a Structured Data Transformer Ensemble for 30-Day AMI Readmission Prediction
This study introduces and temporally validates the Marqi Index, a transformer-based ensemble model that achieves a 0.807 AUROC for predicting 30-day AMI readmissions using only deidentified, structured EHR and claims data, thereby addressing the lack of benchmarks for this specific data environment and outperforming existing structured-data approaches.