Navigating ECG Signal Forecasting: A Systematic Review of Current Trends and Future Directions

Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide, highlighting the urgent need for effective early detection strategies. The electrocardiogram (ECG), as the gold standard for cardiac monitoring, provides critical data for clinical decision-making. Short-term ECG forecasting can support timely detection of abnormal cardiac events, enabling proactive interventions. This systematic literature review (SLR) examines ECG forecasting techniques based on time series analysis, addressing five research questions (RQ1–RQ5) regarding data sources, forecasting models, performance metrics, challenges, and methodological trends. Three databases—PubMed, IEEE Xplore, and ScienceDirect—were systematically searched for peer-reviewed articles published between 2013 and 2023. Following the PRISMA 2020 guidelines and the application of predefined eligibility criteria, eight studies were included in the final synthesis. The analysis reveals that public databases are the preferred source due to accessibility; the results also suggest that hybrid forecasting models dominate current research, preprocessing and analytical approaches vary widely, and performance is primarily evaluated using RMSE and MAE. Key research gaps include limited studies on real-time arrhythmia prediction, lack of standardized evaluation frameworks, and underexploration of hybrid and deep learning strategies in diverse patient populations. Building on these findings, the review proposes a future research agenda (2025–2030) focused on developing automated, real-time ECG forecasting systems with enhanced accuracy and interpretability, leveraging hybrid and deep learning models, and establishing standardized benchmarking protocols. Overall, ECG forecasting is identified as a promising yet underexplored field, offering substantial opportunities for innovation in predictive cardiology and clinical decision support.