Artificial Intelligence and Deep Learning Models Applied to Electrocardiographic Analysis for Cardiac Event Prediction and Clinical Prognosis: A Scoping Review

Cardiac events, including in-hospital cardiac arrest and malignant arrhythmias, remain a major cause of morbidity and mortality worldwide despite advances in resuscitation science and critical care. Evidence suggests that most adult patients who experience cardiac arrest undergo measurable clinical deterioration hours before the event; however, current early warning systems and conventional risk stratification tools show limited sensitivity and contextual adaptability. In recent years, the increasing availability of digital electrocardiographic (ECG) data and advances in artificial intelligence (AI), particularly machine learning and deep learning techniques, have enabled the development of predictive models capable of identifying complex, non-linear patterns that are not readily detectable through traditional clinical interpretation. The purpose of this study is to systematically map and synthesize the existing evidence on the use of artificial intelligence and deep learning models applied to electrocardiographic analysis for the prediction of cardiac events and clinical prognosis in adult populations. Using a scoping review methodology, this project aims to provide a comprehensive overview of how AI-based ECG models have been developed, validated, and applied across different clinical contexts, including emergency departments, hospital wards, intensive care units, and ambulatory or continuous monitoring settings. This scoping review will be conducted in accordance with the Joanna Briggs Institute (JBI) methodological framework and reported following the PRISMA-ScR guidelines. The review will include studies published between 2015 and 2025 that evaluate AI-based approaches using ECG data to predict adverse cardiac outcomes, such as cardiac arrest, major arrhythmias, or other clinically relevant prognostic endpoints in adults. Rather than assessing effectiveness or comparative superiority, the review will focus on mapping the characteristics of the available evidence, including model architectures, ECG data sources, target outcomes, validation strategies, and reported predictive performance metrics. The expected outcomes of this project include (1) a structured synthesis of AI and deep learning methodologies applied to ECG-based cardiac risk prediction, (2) identification of prevailing clinical applications and methodological trends, (3) recognition of limitations and gaps in the current literature, and (4) clarification of areas requiring further research and validation prior to widespread clinical implementation. By organizing a rapidly expanding and methodologically heterogeneous body of literature, this review aims to support future research, inform clinical innovation, and provide a conceptual foundation for the responsible integration of AI-driven ECG analysis into adult cardiovascular care.