# Serial 12-Lead Electrocardiogram–Based Deep-Learning Model for Hospital Admission Prediction in Emergency Department Cardiac Presentations: Retrospective Cohort Study

## TL;DR

A deep-learning model using serial ECGs and clinical data accurately predicts hospital admission for cardiac patients in the emergency department early in their visit.

## Contribution

A novel multimodal deep-learning model that integrates serial ECGs with sequential vitals and clinical data for early hospital admission prediction in ED cardiac patients.

## Key findings

- The multimodal model achieved an AUROC of 0.911 in predicting hospital admission using data up to the last ECG.
- The model predicted disposition a median of 0.3 hours after triage and 4.6 hours before ED departure.
- The multimodal model outperformed baseline models in both single and multiple ECG cohorts.

## Abstract

Emergency department (ED) crowding is often attributed to a slow hospitalization process, leading to reduced quality of care. Predicting early disposition in patients presenting with cardiac issues is challenging: most are ultimately discharged, yet those with a cardiac etiology frequently require hospital admission. Existing scores rely on single-time-point data and often underperform when patient risk evolves during the visit.

This study aimed to develop and validate a real-time deep-learning model that fuses serial 12-lead electrocardiogram (ECG) waveforms with sequential vitals and routinely available clinical data to predict hospital admission early in ED encounters.

We conducted a retrospective cohort study using the Medical Information Mart for Intensive Care (MIMIC) IV, MIMIC-IV Emergency Department module, and MIMIC-IV electrocardiogram module databases. Adults presenting with chest pain, dyspnea, syncope, or presyncope and at least 1 ECG within their ED stay were included. Two evaluation cohorts were defined: all stays with ≥1 ECG (n=30,421) and a subset with ≥2 ECGs during the encounter (n=11,273). To predict hospital admission, we first established 2 baseline models: a tabular model (random forest [RF]) trained on structured clinical variables, including demographics, triage acuity, past medical history, medications, and laboratory results, and an ECG-only model that learned directly from raw 12-lead waveforms. We then developed a multimodal deep-learning model that combined ECGs with sequential vital signs as well as the same static tabular features. All models were restricted to data available during the stay up to the time of the last ECG. Performance was assessed with stratified 5-fold cross-validation using identical splits across models.

The multimodal model achieved an area under receiver operating characteristic (AUROC) of 0.911 when trained on all eligible stays. The model predicted disposition after the final ECG was taken, which was a median of 0.3 (IQR 0.2‐5.3) hours after triage and 4.6 (IQR 2.7‐7.3) hours before ED departure. Baseline models performed worse: the ECG-only model had an AUROC of 0.852, and the tabular RF had an AUROC of 0.886. In the subset requiring at least 2 ECGs within the stay, ECG-only reached an AUROC of 0.859, and RF, with the longer interval to chart tabular data, reached a higher AUROC of 0.911. The multimodal model had an AUROC of 0.924 and outperformed baselines in each cohort (paired DeLong P<.001).

Serial ECGs, when integrated with evolving vitals and routine clinical features, enable accurate, early prediction of ED disposition in patients presenting with cardiac issues. This open-source, reproducible framework highlights the potential of multimodal deep learning to streamline ED flow, prioritize higher risk cases, and detect evolving, time-critical pathology.