Machine Learning Identification of Imminent Sudden Cardiac Arrest Warning Sign Combinations

Article Impact Level: HIGH
Data Quality: STRONG
Summary of  Circulation: Arrhythmia and Electrophysiology,  https://doi.org/10.1161/CIRCEP.125.014647 
Dr. Kyndaron Reinier  et al.

Points

Investigators at Cedars-Sinai evaluated novel risk-prediction strategies for out-of-hospital sudden cardiac arrest, an event affecting over three hundred fifty thousand individuals in the United States annually.
Machine learning analysis of three hundred sixty-four sudden cardiac arrest cases demonstrated that combining warning symptoms like shortness of breath with clinical history predicts imminent arrest.
Sex-specific symptom patterns revealed that chest pain combined with coronary artery disease predicted imminent cardiac arrest in women, whereas chest pain with heart failure predicted it in men.
Longitudinal tracking of over nine thousand hospitalized patients showed that a second acute coronary event tripled cardiac arrest risk, while recurrent heart failure hospitalizations doubled overall risk.
The study authors concluded that integrating acute warning symptom algorithms with recurrent cardiovascular event tracking significantly improves cardiac arrest identification beyond traditional ejection fraction measurements.

Summary

These two observational studies led by Cedars-Sinai investigators evaluated novel risk stratifiers for imminent and long-term out-of-hospital sudden cardiac arrest (SCA). Given that more than two-thirds of the 350,000 annual U.S. out-of-hospital SCA cases occur in individuals without low left ventricular ejection fraction, conventional screening metrics remain insufficient. The research sought to determine if integrating acute warning symptoms, medical histories, and recurrent cardiovascular events could enhance predictive accuracy for imminent SCA (ISCA) and long-term risk.

In the first study (Circulation: Arrhythmia and Electrophysiology), machine learning algorithms analyzed 364 patients with EMS activations who survived ISCA alongside 313 symptomatic non-SCA controls from Oregon and Ventura County, California. Shortness of breath combined with pre-existing coronary artery disease (CAD) or heart failure (HF), as well as isolated seizure-like symptoms, strongly predicted ISCA. Sex-specific interactions revealed that chest pain with CAD predicted ISCA in females, whereas chest pain with HF predicted ISCA in males. Most warning symptoms presented at least 15 minutes before arrest, contrasting with historical cohort data showing that 81% of patients delayed contacting emergency services.

In the second study (Journal of the American Heart Association), investigators tracked 6,700 patients hospitalized for HF and 2,900 patients hospitalized for acute coronary syndrome within the 400,000-resident O.S.C.A.R. cohort. Recurrent cardiovascular events significantly elevated long-term SCA risk: a second acute coronary syndrome event yielded a greater than three-fold increase in SCA risk, while a second HF hospitalization nearly doubled the risk, with risk compounding upon subsequent admissions. The findings demonstrate that combining acute symptom-history profiles with longitudinal disease recurrence markers provides a dual-framework approach to overcome traditional roadblocks in SCA risk prediction.

Link to the article: https://www.ahajournals.org/doi/10.1161/CIRCEP.125.014647

References

Reinier, K., Chugh, H., Kadiyala, V., Sargsyan, A., Uy-Evanado, A., Nakamura, K., Heckard, E., Mathias, M., Grogan, T., Elashoff, D., Salvucci, A., Jui, J., & Chugh, S. S. (2026). Clinical triage of individuals with warning symptoms of imminent cardiac arrest. Circulation: Arrhythmia and Electrophysiology, e014647. https://doi.org/10.1161/CIRCEP.125.014647