Open research questions in ECG Monitoring and Analysis

What the literature leaves open

• The need for a holistic framework for ECG arrhythmia classification. The requirement for a clear and systematic approach to ECG arrhythmia classification. The importance of achieving high accuracy in ECG arrhythmia classification.

  Machine learning–enabled ECG arrhythmia classification: a systematic and educational study from signal processing to decision support · 2026 · DOI

• Measuring cardiac quantitative data manually on point-of-care ultrasonography images can be time-consuming and subject to user experience. Artificial intelligence can potentially automate this quantification but requires evaluation. Prior studies have used images from comprehensive echocardiography examinations obtained in echocardiography laboratories.

  Automated artificial intelligence–enabled measurement of cardiac structures on point-of-care ultrasonography: a prospective multicenter validation study · 2026 · DOI

• in adults: from 28. Mercaldo SF, Hillis JM, Blume JD. Evaluating the performance and clinical utility of AI-driven diagnostic tools in radiology. Radiology. (2025) 317(2):e243935. doi: 10. 1148/radiol.243935 29. Chai T, Draxler RR. Root mean square error (RMSE) or mean absolute error (MAE)? – arguments against avoiding RMSE in the literature. Geosci Model Dev. (2014) 7(3):1247–50. doi: 10.5194/gmd-7-1247-2014 30. Barnhart HX, Kosinski AS, Haber MJ. Assessing individual agreement. J Biopharm Stat. (2007) 17(4):697–719. doi: 10.1080/10543400701329489 31. Martin Bland J, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. (1986) 1(8476):307–10. doi: 10.1016/ S0140-6736(86)90837-8 32. Bartko JJ. The intraclass correlation coefficient as a measure of reliability. Psychol Rep. (1966) 19(1):3–11. doi: 10.2466/pr0.1966.19.1.3 33. Jing J, Sun H, Kim JA, Herlopian A, Karakis I, Ng M, et al. Development of during of expert-level electroencephalogram interpretation. JAMA Neurol. (2020) 77(1):103–8. doi: 10. 1001/jamaneurol.2019.3485 epileptiform automated discharges detection 34. Furtado S, Reis L. Inferior vena cava evaluation in fluid therapy decision making in intensive care: practical implications. Rev Bras Ter Intensiva. (2019) 31(2):240–7. Avaliacao da veia cava inferior na decisao de fluidoterapia em cuidados intensivos: implicacoes praticas. doi: 10.5935/0103-507X.20190039 35. Miller A, Mandeville J. Predicting and measuring fluid responsiveness with echocardiography. Echo Res Pract. (2016) 3(2):G1–2. doi: 10.1530/ERP-16-0008

  Automated artificial intelligence–enabled measurement of cardiac structures on point-of-care ultrasonography: a prospective multicenter validation study · 2026 · DOI

• Investigating the model's performance in different patient populations, - Developing strategies to improve the model's performance in patients with prior heart disease, - Exploring the use of the model in combination with other diagnostic tools, - Evaluating the model's performance in real-world clinical settings

  Training and validation of a 12-lead ECG-based deep-learning model for myocardial infarction subtypes · 2026 · DOI

• There is an unmet clinical need for automated, reliable, and objective ECG interpretation. Manual interpretation of 12-lead ECGs relies heavily on reader expertise and exhibits substantial inter-reader variability.

  Training and validation of a 12-lead ECG-based deep-learning model for myocardial infarction subtypes · 2026 · DOI

• The integration of digital health technologies into clinical practice poses challenges for healthcare providers, including lack of knowledge and confidence in using the devices. There are also challenges related to patient access, socioeconomic disparities, and cybersecurity threats. The paper identifies the need for clear guidance on legal obligations and reimbursement proceedings as a significant challenge.

  Artificial intelligence enabled mobile health technologies in arrhythmias-an opinion article on recent findings · 2025 · DOI

• Lack of knowledge about the features of the devices, - Lack of confidence in the use of devices, - Worries about liability, - Socioeconomic disparities that impact access, - Potential anxiety related to test results, - Differing levels of digital literacy

  Artificial intelligence enabled mobile health technologies in arrhythmias-an opinion article on recent findings · 2025 · DOI

• Class imbalance in the dataset, addressed by capping the maximum number of samples per class. Data scarcity for certain arrhythmias, such as Wolff-Parkinson-White syndrome. Need for automated diagnostic tools that can support clinicians and reduce human error.

  ECG-XPLAIM: eXPlainable Locally-adaptive Artificial Intelligence Model for arrhythmia detection from large-scale electrocardiogram data · 2025 · DOI

• Data scarcity necessitated a 1:2 positive-to-negative ratio for WPW detection, - Class imbalance was addressed by capping the maximum number of samples per class at 50,000, - The model was trained on the MIMIC-IV dataset and externally validated on PTB-XL, - Limited information is provided about the model's performance on other datasets

  ECG-XPLAIM: eXPlainable Locally-adaptive Artificial Intelligence Model for arrhythmia detection from large-scale electrocardiogram data · 2025 · DOI

• However, training accurate delineation models requires manual annotations that are scarce and time-consuming to obtain.

  Decoder Design Matters for ECG Delineation · 2026

• While recent SR methods based on discriminative neural networks have shown strong performance, their robustness under distribution shift remains uncertain, which poses a key challenge for real-world deployment.

  DiSR-ECG: Residual Shifting Conditional Diffusion for Robust ECG Super-Resolution · 2026

• However, current evaluation rarely reflects the full capabilities expected of ECG foundation models, with prior work often using narrow task selections and inconsistent datasets, hindering fair comparison.

  BenchECG and xECG: a benchmark and baseline for ECG foundation models · 2026 · DOI

• However, current evaluation rarely reflects the full capabilities expected of ECG foundation models, with prior work often using narrow task selections and inconsistent datasets, hindering fair comparison.

  BenchECG and xECG: a benchmark and baseline for ECG foundation models · 2026 · DOI

• Background: The impact of artificial intelligence in improving Tele-ECG response times and diagnostic accuracy among emergency patients experiencing acute chest pain remains uncertain.

  Evaluation of AI-enhanced tele-ECG response time and diagnosis in acute chest pain patients · 2025 · DOI

• However, the ability of AI to identify abnormalities from single-lead recordings across a range of pathological conditions remains to be systematically investigated.

  Convolutional neural network (CNN)-enabled electrocardiogram (ECG) analysis: a comparison between standard twelve-lead and single-lead setups · 2024 · DOI

• Although deep learning based methods using segmentation models to locate P, QRS, and T waves have shown promising results, their ability to handle arrhythmias has not been studied in any detail.

  Deep learning based ECG segmentation for delineation of diverse arrhythmias · 2024 · DOI

• The study identifies the challenge of deploying complex models on resource-constrained edge devices. The challenge of achieving a balance between classification performance and real-time deployability is also noted. The need for low-cost and real-time AFib detection systems is a significant challenge.

  Modeling and Evaluating an Intelligent Health Monitoring System for Detecting Atrial Fibrillation · 2026 · DOI

• The study does not provide a comprehensive comparison with other existing AFib detection systems. The evaluation is limited to a specific dataset and may not generalize to other populations or environments.

  Modeling and Evaluating an Intelligent Health Monitoring System for Detecting Atrial Fibrillation · 2026 · DOI

• Conventional cloud-based health monitoring solutions suffer from limitations such as high energy usage, communication delays, and privacy concerns. Conventional monitoring approaches rely heavily on periodic clinical visits or cloud-based telemedicine systems, which introduce latency and increase energy consumption.

  Edge-Based Tinyml Framework for Intelligent Cardiac Drug Response Monitoring Using Embedded Hardware Software Co-Design · 2026 · DOI

• and hyperparameter tuning. This proactive optimization avoided excessive computational redundancy and ensured deterministic execution. The outcome demonstrates that embedded microcontrollers, when paired with tailored TinyML models, can perform sophisticated biomedical architecture alterations selection during analytics without infrastructure.

  Edge-Based Tinyml Framework for Intelligent Cardiac Drug Response Monitoring Using Embedded Hardware Software Co-Design · 2026 · DOI

• Maintaining diagnostic accuracy while achieving high compression remains difficult. Algorithms must operate efficiently with minimal delay. Wearable devices require low-power algorithms.

  A Detailed Survey of Electrocardiogram Signal Reduction Strategies: Evolutions, Hurdles, and Perspectives · 2026 · DOI

• The need for effective ECG compression has intensified with the rise of telehealth and long-term cardiac surveillance via wearables. Maintaining diagnostic accuracy while achieving high compression remains difficult.

  A Detailed Survey of Electrocardiogram Signal Reduction Strategies: Evolutions, Hurdles, and Perspectives · 2026 · DOI

• The sample size was limited to 13 standard 12-lead ECGs, - The study was an exploratory evaluation, - The number of cases per category was small, which would yield unstable and potentially misleading estimates, - No application programming interface (API) parameters were modified, - No system prompts beyond the standardized evaluation prompt were applied

  Artificial Intelligence for Biomedical Diagnostics: Diagnostic Accuracy and Reliability of Multimodal Large Language Models in Electrocardiogram Interpretation · 2026 · DOI

• Larger studies are needed to confirm the findings, - Task-specific validation of general-purpose MLLMs is required before deployment, - Real-world clinical assessment of AI systems in medicine is necessary prior to implementation, - Further research on the use of MLLMs in structured screening scenarios is needed

  Artificial Intelligence for Biomedical Diagnostics: Diagnostic Accuracy and Reliability of Multimodal Large Language Models in Electrocardiogram Interpretation · 2026 · DOI

• Despite its advantages, the proposed system has certain limitations that need to be considered. One of the primary challenges is the requirement for a large and diverse dataset. Deep learning models generally perform better when trained on extensive datasets, and limited data can affect the model’s ability to generalize across different patient conditions. Another limitation is the high computational cost associated with training deep learning models. The use of multiple architectures, especially hybrid models, requires significant processing power and memory, which may not be readily available in all environments. The current system is also limited to ECG image-based classification, which means it does not directly process raw ECG signal data or multi- lead inputs. This restricts its applicability in more advanced clinical scenarios. Furthermore, the system is not yet clinically deployed, meaning it still requires validation and approval before being used in real-world healthcare settings. 7. Future Enhancement: To overcome the current limitations and improve the system further, several enhancements can be considered in future work. One important direction is the integration of multi-lead ECG data, which can provide more comprehensive information about cardiac activity and improve classification, accuracy. Another promising enhancement is cloud- based deployment, which would allow the system to be accused remotedly and handle large- scale data efficiently. © 2026 The Author(s). Published by IJCOPE Journal. Website: https://ijcope.org/ 5 International Journal of Creative and Open Research in Engineering and Management ISSN: 3108-1754 (Online) Volume 02 Issue 04 April-2026 | Impact Factor: 3.5 This would also reduce the need for high-end local hardware. Additionally, developing a mobile application can make the system more accessible to healthcare professionals, especially in remote or rural areas. In corporating Explainable AI (XAI) techniques is another crucial improvement. This would allow the system to provide clear explanations for its predictions, increasing trust and transparency among medical practitioners. Finally, integrating the system with hospital management systems can streamline workflows and enable seamless data sharing, making the solution more practical for real-world applications. 8. Conclusion: In conclusion, this research demonstrates the effectiveness of deep learning techniques in the classification of ECG signals for detecting heart diseases. By comparing multiple models, including CNN, MobileNet and DenseNet, the study highlights the superior performance of the hybrid MobileNet + LSTM model. This model successfully combines spatial features extraction with temporal pattern recognition, resulting in improved accuracy and reliability. The proposed system not only enhances diagnostic accuracy but also reduces the burden on healthcare professionals by automating the analysis process. Its ability to deliver fast and consistent results makes it a valuable tool for supporting clinical decision-making. Although there are certain limitations, the system shows strong potential for real- world implementation with further improvements and validation. Overall this work represents a meaningful step toward the development of intelligent healthcare solutions, where artificial intelligence can assist doctors in providing faster, more accurate, and more efficient patient care.

  Automated Heart Disease Detection from ECG Signals using a Hybrid Deep Learning Approach · 2026 · DOI

Most-cited papers in ECG Monitoring and Analysis

• An artificial intelligence-enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction · The Lancet · 2019 · 1,510 citations
• How to use digital devices to detect and manage arrhythmias: an EHRA practical guide · EP Europace · 2022 · 230 citations
• Detection of Cardiovascular Diseases in ECG Images Using Machine Learning and Deep Learning Methods · IEEE Transactions on Artificial Intelligence · 2022 · 224 citations
• A transformer-based deep neural network for arrhythmia detection using continuous ECG signals · Computers in Biology and Medicine · 2022 · 220 citations
• Classifying Cardiac Arrhythmia from ECG Signal Using 1D CNN Deep Learning Model · Mathematics · 2023 · 187 citations
• Analysis of various techniques for ECG signal in healthcare, past, present, and future · Biomedical Engineering Advances · 2023 · 166 citations
• Usefulness of Automated Serial 12-Lead ECG Monitoring During the Initial Emergency Department Evaluation of Patients With Chest Pain · Annals of Emergency Medicine · 1998 · 152 citations
• An efficient and robust Phonocardiography (PCG)-based Valvular Heart Diseases (VHD) detection framework using Vision Transformer (ViT) · Computers in Biology and Medicine · 2023 · 114 citations
• Artificial intelligence-enhanced electrocardiography for accurate diagnosis and management of cardiovascular diseases · Journal of Electrocardiology · 2024 · 103 citations
• CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification · Biomedical Signal Processing and Control · 2024 · 99 citations

Most recent work

• Teaching multimodal LLMs to comprehend 12-lead electrocardiographic images · npj Digital Medicine · 2026
• An ECG biomarker for sudden cardiac death discovered with deep learning · Nature · 2026
• Wearable device derived electrocardiographic age and its association with atrial fibrillation · npj Digital Medicine · 2026
• Digitizing paper ECGs at scale: an open-source algorithm for clinical research · npj Digital Medicine · 2026
• Leipzig Heart Center ECG-database: Arrhythmias in children and patients with congenital heart disease · Physiological Measurement · 2026
• Uncertainty Quantification in Machine Learning for Biosignal Applications - A Review · Journal of Healthcare Informatics Research · 2026
• Generative AI for ECG Interpretation Education: Impact on Nursing, Student Performance, and AI Model Accuracy · Nurse Educator · 2026
• xGNN4MI: explainability of graph neural networks in 12-lead electrocardiography for cardiovascular disease classification · npj Digital Medicine · 2026
• Performance of the 12-lead ECG in predicting short- and long-term risk of sudden cardiac death · npj Digital Medicine · 2026
• External validation of ECG artificial intelligence for emergency and cardiac assessment across a large-scale U.S. healthcare system · npj Digital Medicine · 2026

Related topics in Medicine