Background: National electronic health record (EHR) networks can support learning health systems (LHSs) by enabling large-scale data aggregation, monitoring, and benchmarking, but their capacity to pr

The gap

Consensus across the literature

Supporting evidence — 5 representative gaps

• FEDERATED AND EXPLAINABLE ARTIFICIAL INTELLIGENCE FOR PRIVACY-PRESERVING CLINICAL DECISION SUPPORT SYSTEMS (2026) · doi

  Although the proposed framework shows promising results in combining Federated Learning (FL) with Explainable AI (XAI), there are still several areas where the system can be further improved. Future research can focus on enhancing performance, adaptability, and real-world applicability of the model. X.I. Personalized Federated Learning (pFL): One of the main challenges in federated learning is data heterogeneity, meaning that data collected from different hospitals may vary significantly in terms of quality, distribution, and patient demographics.In the current system, a single global model is shared across all clients. However, this approach may not always perform equally well for every institution. To address this issue, future work can explore Personalized Federated Learning (pFL).In this approach, instead of using one common model, each client can have a slightly customized version of the global model that better fits its local data. This can improve accuracy and make the system more adaptable to different healthcare environments. 554 International Journal of Advance and Innovative Research Volume 13, Issue 2: April - June 2026 ISSN 2394 - 7780 X.II. Integration of Multi-Modal Data: At present, the system mainly focuses on specific types of data such as medical images or structured health records. However, real-world healthcare data is much more complex and comes in different forms. Future improvements can include the integration of multi-modal data, such as: Electronic Health Records (EHRs) Medical imaging (X-rays, MRIs, CT scans)

  Keywords: federated learning system model future different real world personalized global approach issue healthcare integration multi
• Artificial Intelligence in Cardiovascular Imaging: Current Landscape, Clinical Impact, and Future Directions (2025) · doi

  8.1 Emerging trends in cardiovascular imaging Healthcare is becoming more personalized and precise due to the development of AI technologies and their integration into various fields of medicine, such as cardiovascular imaging. This will enhance patient outcomes by providing early diagnosis and prompt management of CVD120. Although AI models have numerous advantages, some limitations prevent them from being widely used in medicine. One of them is that most of the AI models are trained on data that is small, narrow, and not diverse. This could result in lower generalizability and over-fitting. To overcome this issue, Federated Learning (FL) or Transfer Learning methods can be used, which train the models with data from different sites without centralising it. This leads to a decreased risk of data leaks or unauthorized access as the data is not sent over to other networks121,122. Due to training with multiple datasets, the model becomes familiar with rare cases, which increases its sensitivity and results in decisions that are less biased. The performance of such models is similar to those trained on centrally hosted datasets and much better when compared to models trained with data from a single institution, according to the latest studies123. Another disadvantage of AI models is their “unexplainable” feature, also known as “black- boxes”. This means that how the AI reaches a specific conclusion or diagnosis is not known. The lack of transparency its working124,125. Explainable AI methods are being developed to resolve this issue126. Class Activation Mapping (CAM) is a technique that comes under Explainable AI. CAM can identify the input patterns within the deep neural network, which leads to the activation of certain outputs and shows the findings on the image. This helps the physicians who use AI to better understand the reasoning of the model and make informed decisions127. AI can also serve as a means of bringing together different imaging modalities, which can be extremely helpful, especially in the diagnosis of heterogeneous regarding questions raises diseases, for example, heart failure and atrial fibrillation128. The multimodality AI approach combines information, in the form of image, text, audio, video, and language, obtained from different imaging methods. Each modality offers important additional data, resulting in accurate outputs. A study was done to distinguish between the causes of left ventricular hypertrophy by merging data from ECG and echocardiography. The multimodal AI was found to have higher sensitivity and specificity compared to physicians129. Therefore, as AI technologies are evolving rapidly, it is changing how cardiovascular imaging is used, by improving image quality, reducing the time and workforce required for image analysis, and providing prognostic information10,130. 8.2 Wearable AI Devices timely detection ECG is an essential and most frequently used non- invasive test in cardiology. However, to record and interpret an ECG, the individual is required to visit a healthcare professional. This results in paroxysmal arrhythmias being undetected. The integration of ECG monitoring and AI-based ECG interpretation into wearable devices like smartwatches has led to widespread access to ECG131. Furthermore, there has been an increase in the detection of arrhythmias131,132. In case of AF, early detection is crucial to avoid progression and development of complications such as stroke. As most cases have either nonspecific symptoms or are asymptomatic, patients are unaware of their condition and hence do not seek medical care. Using wearable devices with AI technology, early and in improved treatment outcomes133. Apart from arrhythmias, such devices can also be helpful in the monitoring of patients with other CVDs, for example, CHD, Left Ventricular Systolic

  Keywords: models imaging used image devices cardiovascular early diagnosis trained different wearable detection arrhythmias healthcare development
• Artificial Intelligence in Cardiovascular Imaging: Current Landscape, Clinical Impact, and Future Directions (2025) · doi

  ● Generalizability to the population as a whole: How can federated and transfer learning be utilized to achieve robust performance for individuals in a range of ethnic, age, and comorbid disease groups without compromising data sharing121-123? ● Trade-off between explainability and performance: How much transparency of the model (XAI techniques, CAM maps) is needed regulatory agency for clinician approval, and what implications does this have for clinical accuracy diagnostic implementation? trust and and www.discoveriesjournals.org/discoveries 19 AI in Cardiovascular Imaging: State and Outlook ● Integrating into the clinical workflow: What are the optimal AI-human interaction paradigms (e.g., real-time triage, HITL oversight, automated reporting) to maximize efficiency gains without trading off safety in high-acuity settings? ● Regulation and liability: What processes should be used to “oversee” dynamic “learning” algorithms in operation, to handle degradation, to ensure patient safety, and to apportion liability among and practitioners? the developers, organisations, economic ● Cost-effectiveness and access: What will be the of AI sustained implementation on health-system costs, and how can AI solutions be scaled resource- constrained settings to mitigate inequalities? implications to

  Keywords: learning performance without implications clinical implementation safety settings liability generalizability population whole federated transfer utilized
• COMPARATIVE PERFORMANCE ANALYSIS OF MACHINE LEARNING AND ARTIFICIAL NEURAL NETWORK MODELS FOR HEART DISEASE FORECASTING (2026) · doi

  In Future, this work can be better by applying k fold cross validation, developing real time healthcare decision support applications, using explainable AI tools such as SHAP or LIME, exploring hybrid deep learning and ensemble models,using some external validation data, and testing on larger multi center datasets . REFERENCES R. Detrano et al., “International application of a the new probability algorithm diagnosis of coronary artery disease,” American Journal of Cardiology, vol. 64, no. 5, pp. 304–310, 1989. for C. Cortes and V. Vapnik, “Support-vector networks,” Machine Learning, vol. 20, no. 3, pp. 273–297, 1995. R. C. Deo, “Machine learning in medicine,” Circulation, vol. 132, no. 20, pp. 1920– 1930, 2015. L. Breiman, A. Rajkomar, J. Dean, and I. Kohane, “Machine learning in medicine,” New England Journal of Medicine, vol. 380, no. 14, pp. 1347–1358, 2019. “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. S. B. Kotsiantis, “Supervised machine learning: A review of classification techniques,” Informatica, vol. 31, pp. 249–268, 2007. H. K. Bashir, M. Mansoor, and A. R. Javed, “Machine learning-based heart disease prediction system,” Diagnostics, vol. 11, no. 5, 2021. T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” in Proceedings of the 22nd ACM SIGKDD International

  Keywords: learning machine medicine validation support using international disease journal system future better applying fold cross
• Opportunities and Challenges in Using National EHR Networks for AI in Learning Health Systems (2026) · doi

  Background: National electronic health record (EHR) networks can support learning health systems (LHSs) by enabling large-scale data aggregation, monitoring, and benchmarking, but their capacity to produce trustworthy and locally deployable machine learning and artificial intelligence (ML/AI) models remains uncertain.

  Keywords: health learning background national electronic record networks support systems lhss enabling large scale aggregation monitoring