Journal DOI 10.68956/journal.human-performance-futures

Creative Commons source reading · Machine learning for performance

Artificial intelligence applied to post-resuscitation ECGs for early prognostication after out-of-hospital cardiac arrest

Gentile FR; Shah STH; Sperti M; Panagiotopoulos K; Primi R; Ulmanova L; Currao A; Bendotti S; Baldi E; Bauer DN; Pontremoli SM; Marconi G; Deriu MA; Savastano S

Originally published in Front Cardiovasc Med · 2026-05-11

From the article

Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.

Abstract

AbstractBackgroundThere are limited tools for early outcome prediction after Out-of-Hospital Cardiac Arrest (OHCA). This study aimed to evaluate whether a machine learning model could help to predict neurological outcome using the 12-lead ECG acquired on scene following return of spontaneous circulation (ROSC).MethodsWe conducted a retrospective analysis of prospectively collected post-ROSC ECGs from the Lombardy Cardiac Arrest Registry (January 2015-December 2023). The study included all the patients resuscitated from OHCA with a 12-lead ECG acquired on scene after ROSC. A deep neural network model was developed, validated, and tested using features extracted from these ECGs via computer vision techniques, incorporating patient age, sex, initial rhythm, and ROSC-ECG time. The dataset was split into training (80%), validation (10%),…

Discussion

To the best of our knowledge, this study presents one of the first machine learning approaches leveraging post-ROSC 12-lead ECG data to evaluate the association between ECG-derived features and neurological outcome immediately after ROSC. We have developed and internally validated a machine-learning algorithm intended to support the assessment of neurological outcome at hospital discharge by using post-ROSC 12-lead ECG together with few other variables in patients resuscitated from OHCA. This model demonstrates that integrating ECG-derived features may improve early risk stratification at the scene, prior to hospital admission. Besides its performance, a prominent distinction of our tool is its reliance on patient's individual data, namely the post-ROSC ECG, which is readily obtainable by emergency personnel and recommended by both…

Conclusion

To the best of our knowledge, this is the first machine-learning tool based on post-ROSC 12-lead ECGs designed to contribute to the assessment of neurological outcome in the pre-hospital setting. Our DNN model is intended as an easily applicable tool that may support early stratification of patients' prognosis after an out-of-hospital cardiac arrest, therefore supporting EMS personnel in their decision-making process about ACLS treatment, triage and transport. Further external validation is required to determine the generalizability and clinical utility of this tool across diverse healthcare systems, particularly in light of variability in EMS organization and practice worldwide.

Editorial reading guide

  • What data created and tested the model, and was evaluation independent?
  • What is the cost of a false positive, false negative or out-of-distribution decision?