Key result
A multimodal deep learning model integrating raw ECG waveforms and clinical routine data achieved an AUROC of 0.9115 for predicting patient deterioration, outperforming clinical routine data alone.
Why the study?
The integration of multimodal data, including raw waveform signals, remains underexplored in clinical decision support for emergency care.
Does a multimodal AI model incorporating raw ECG waveforms improve the prediction of discharge diagnoses and patient deterioration in emergency department patients compared to models using only clinical routine data or ECG features?
Observational (n=71,098)
No
Does a multimodal AI model incorporating raw ECG waveforms improve the prediction of discharge diagnoses and patient deterioration in emergency department patients compared to models using only clinical routine data or ECG features?
Incorporating raw ECG waveforms into multimodal AI decision support models significantly improves the prediction of discharge diagnoses and clinical deterioration in emergency department patients.
May support raw ECG integration into ED deterioration models; hypothesis-generating and requires prospective validation before practice change.
BACKGROUND: AI-driven prediction algorithms have the potential to enhance emergency medicine by enabling rapid and accurate decision-making regarding patient status and potential deterioration. However, the integration of multimodal data, including raw waveform signals, remains underexplored in clinical decision support. METHODS: We present a dataset and benchmarking protocol designed to advance multimodal decision support in emergency care. Our models utilize demographics, biometrics, vital signs, laboratory values, and electrocardiogram (ECG) waveforms as inputs to predict both discharge diagnoses and patient deterioration. RESULTS: The diagnostic model achieves area under the receiver operating curve (AUROC) scores above 0.8 for 609 out of 1,428 conditions, covering both cardiac (e.g., myocardial infarction) and non-cardiac (e.g., renal disease, diabetes) diagnoses. The deterioration model attains AUROC scores above 0.8 for 14 out of 15 targets, accurately predicting critical events such as cardiac arrest, mechanical ventilation, ICU admission, and mortality. CONCLUSIONS: Our study highlights the positive impact of incorporating raw waveform data into decision support models, improving predictive performance. By introducing a unique, publicly available dataset and baseline models, we provide a foundation for measurable progress in AI-driven decision support for emergency care.
No takes yet. Share an insight, caveat, or question.
Alcaraz et al. (2025) conducted an observational in Emergency department patients (n=71,098). Multimodal deep learning model (ECG waveforms + clinical routine data) vs. Unimodal models (clinical routine data alone or ECG features alone) was evaluated on Area under the receiver operating curve (AUROC) for predicting patient deterioration (95% CI 0.8991-0.9222). A multimodal deep learning model integrating raw ECG waveforms and clinical routine data achieved an AUROC of 0.9115 for predicting patient deterioration, outperforming clinical routine data alone.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: