Introduction

Artificial intelligence (AI) is exerting a growing influence on cardiology, likely contributing to improvements in diagnostic accuracy, predictive analytics, and personalized patient management.¹ Initially, AI applications were restricted to automating apparently simple and routine tasks, such as electrocardiogram (ECG) interpretation.² However, the advent of more refined machine learning (ML) algorithms has enabled more sophisticated applications, including advanced image-based diagnostics, complex risk prediction models, and real-time procedural guidance (Figure 1).³

Despite these advancements, AI integration into clinical practice remains constrained by several factors, including the need for regulatory approval, extensive validation, and real-world applicability. Indeed, while many AI algorithms and applications demonstrate high performance in retrospective analyses, they encounter challenges in generalizability when applied to diverse patient populations or challenging settings.^4,5 Indeed, in order to ensure robust clinical adoption, external validation in representative cohorts is essential, particularly in interventional cardiology where real-time decision-making and seamless articulation of materials are paramount.

Core Technologies and Learning Paradigms in Cardiology

Several AI-related technologies have significantly impacted cardiology (Table 1; Figure 1).⁶ First, ML algorithms have already been employed to analyze complex cardiovascular data, enhancing diagnostic accuracy and risk prediction in areas ranging from coronary artery disease to heart failure and arrhythmias (Figure 2).^7,8 Deep learning (DL) architectures have further advanced image and signal analysis in cardiac care, despite increasing analytical complexity and accompanying limits in transparency and explainability.⁹

Different learning paradigms are utilized in cardiology to address diverse challenges.¹⁰ Indeed, supervised learning relies on labeled datasets to train models for specific tasks, such as detecting arrhythmias from ECG data. Unsupervised learning, through clustering techniques or similar approaches that aim at finding insights without external labels or indicators, aids, for instance in patient stratification by identifying patterns without predefined labels.^11-13 Reinforcement learning (RL) has also been explored to optimize clinical decision-making by learning from the outcomes of various treatment strategies.

Hybrid models, which combine elements of ML and DL, have been developed with the specific goal of enabling more complex cardiology applications.¹⁴ These models integrate multiple data sources, including imaging and clinical data, to provide a comprehensive analysis of cardiovascular conditions. Ensemble methods, which aggregate predictions from various models, have been employed to improve predictive accuracy. Such approaches facilitate a more holistic understanding of cardiac conditions, leading to better-informed treatment decisions.

Notably, state-of-the-art AI learning models hold the promise of providing numerous benefits in cardiology at large as well as in interventional cardiology, including enhanced diagnostic accuracy, personalized outcome prediction, and improved procedural precision (Table 2), with AI-driven tools supporting clinical decision-making, data augmentation, and telemedicine integration, thus facilitating more efficient and accessible care.¹⁵ The reader should, however, be aware that many applications of the above-mentioned AI tools remain untested or proven effective only in selected populations.

Big Data and Preprocessing for Artificial Intelligence in Cardiology

In cardiology, the integration of big data will prove seminal for further advancing patient care and research.¹⁶ Large datasets, combining clinical, imaging, and biomarker data, enable AI to possibly identify patterns often missed by traditional models, particularly in niche populations like frail young individuals or highly fit elderly patients (Table 3).¹⁷

Effectiveness of AI relies on high-quality data preprocessing, yet clinical data are often noisy, incomplete, and heterogeneous due to variations in acquisition protocols. Techniques such as normalization, noise reduction, and imputation of missing values improve AI accuracy, as seen in ECG signal filtering for arrhythmia detection.⁸ Challenges, however, remain in standardizing and integrating data from diverse healthcare settings, where inconsistent formats hinder AI model generalizability.¹⁸ Privacy concerns are also critical, requiring adherence to regulations like the General Data Protection Regulation and the Health Insurance Portability and Accountability Act.¹⁹ Moreover, algorithmic bias remains a concern, as underrepresentation of certain populations can lead to disparities in AI-driven decision-making. External validation across diverse cohorts is essential to ensure equitable application.²⁰

Artificial Intelligence in Cardiac Imaging and Diagnostics

Developments in cardiac imaging include enhancing accuracy, efficiency, and automation of major imaging modalities, including echocardiography, computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography.²¹ Novel AI-powered tools, for instance automate chamber measurements in echocardiography, improving reproducibility and diagnostic precision, achieving performance comparable to expert sonographers (Table 3).⁸ In cardiac CT and MRI, AI may quicken image reconstruction, shorten scan times, and enable automated segmentation of cardiac structures for precise anatomical and functional assessment.

Despite these advancements, AI-based imaging tools face challenges regarding generalizability, regulatory approval, and integration into clinical workflows. Most poignantly, many algorithms perform well in controlled settings but require external validation in diverse patient populations to ensure broad applicability.^22,23 Moreover, the role of AI in real-world clinical decision-making remains to be fully established, as most studies focus on retrospective datasets rather than prospective validation in randomized trials. Results of ongoing large-scale clinical trials, such as the TRANSFORM (Treating Atherosclerosis In Patients With No Symptoms For Reducing Myocardial Infarction) study, aiming to evaluate the predictive value of AI-based imaging analysis for cardiovascular events in over 7000 patients, are eagerly awaited.²⁴

Indeed, while AI holds promise in revolutionizing cardiac imaging, ongoing research must focus on ensuring transparency, reducing bias, and improving real-time interpretability. Future advancements should prioritize externally validated AI models that seamlessly integrate into clinical practice, balancing automation with physician oversight to optimize diagnostic accuracy and patient care.

Predictive Modeling and Personalized Treatment

The clinical translation of improved predictive modeling into personalized decision-making remains a challenge.⁵ While ML techniques can identify patterns linked to myocardial infarction, heart failure, and stroke, their reported predictive accuracy rarely translates to real-world settings due to overfitting, biased training data, and poor generalizability.²⁵

Similarly, claims on the favorable impact of AI-guided therapy on drug selection and adherence strategies are reassuring, but evidence of benefits over clinician-guided care remains limited.^1,4 Most AI applications in such settings lack prospective validation, struggle with dynamic patient conditions, and are difficult to integrate into routine workflows.

Focusing more attentively on cardiovascular therapeutics, AI-based decision support tools generate automated alerts and inform procedural steps but remain prone to false positives, inappropriate recommendations, and algorithmic bias.²⁶ In addition, over-reliance on these systems risks introducing medical errors rather than improving outcomes, as well as undermining clinician expertise.

Additional challenges to routine application of AI in cardiology include regulatory uncertainty and lack of transparency/interpretability/explainability. Future efforts should prioritize external validation, bias reduction, and ensuring AI remains a support tool rather than replacing expert judgment.

Artificial Intelligence Applications in Interventional Cardiology

Several AI applications have been envisioned in invasive cardiology (Table 4, Table 5, Figure 3).^27-29 Some have targeted the analysis of complex cardiovascular datasets, particularly in detecting ischemic heart disease and characterizing coronary plaque morphology, yet their reliability outside controlled environments remains questionable. While AI-driven image analysis can aid in identifying vulnerable plaques and assessing lesion severity, real-world validation remains limited and often biased toward retrospective datasets.

In the catheterization laboratory, AI-based tools also offer automated image interpretation during angiography and procedural navigation assistance.²⁷ However, while AI enhances imaging clarity and workflow efficiency, the assumption that AI-driven decision-making can reliably mimic human clinical judgment is controversial. Training models on historical cases may, however, yield suboptimal recommendations in dynamic, high-risk scenarios where operator experience and real-time adaptation are critical. In addition, the risk of misclassification, overreliance, and potential liability concerns remains largely unaddressed.

Beyond procedural support, AI extends to post-procedural monitoring and rehabilitation, enabling automated risk assessment and early detection of complications.³⁰ Predictive analytics could optimize post-discharge management, but their integration into standard practice is constrained by poor interpretability, lack of clinician trust, and insufficient regulatory oversight. Without external validation in diverse populations and well-powered prospective studies, the promise of AI enhancing interventional cardiology remains speculative rather than transformative.

Robotics in Interventional Cardiology

Robotic-assisted percutaneous coronary intervention (R-PCI) is an emerging technology in transcatheter therapeutics (Table 6).²⁹ By allowing interventional cardiologists to manipulate guidewires, catheters, and devices from a nearby or even remote cockpit, R-PCI could reduce radiation exposure and minimize the risk of injuries to physicians.

While industry-backed studies report high procedural success rates, real-world implementation has revealed several critical limitations. The lack of tactile feedback can impede complex lesion navigation, increasing reliance on operator intuition and imaging modalities. Moreover, system setup times can extend procedures, potentially negating efficiency benefits.³¹ Cost is another major barrier, with R-PCI requiring substantial investment in robotic platforms, maintenance, and specialized training—a limiting factor for many hospitals. Patients with complex coronary artery disease (e.g., highly calcific lesions) and those with unstable coronary syndromes (e.g., ST-elevation myocardial infarction complicated by shock) could pose additional and quite significant challenges, for instance for inability to maneuver guidewires efficiently via the robot or delays in setting up the system or switching from R-PCI to manual PCI.

The integration of AI with R-PCI has been proposed to enhance real-time decision support and automate procedural adjustments, but its effectiveness remains largely theoretical. No robust, independent studies confirm AI-driven R-PCI improves outcomes over experienced manual operators.³² Furthermore, automation raises concerns over accountability—should AI-driven errors lead to complications, who bears responsibility: the physician, the hospital, or the AI developer?

Ethical, Regulatory, and Collaborative Aspects of Artificial Intelligence in Cardiology

Implementing AI in cardiology introduces several ethical and legal challenges that must be addressed to ensure patient safety and trust (Figure 4).³³ Concerns regarding patient confidentiality and data security are paramount, as AI systems often require access to sensitive health information (Table 3). Additionally, the potential for algorithmic bias poses risks of unequal treatment outcomes across diverse patient populations. Establishing clear legal frameworks and ethical guidelines is essential to navigate these complexities and promote equitable AI integration in cardiovascular care.

Regulatory oversight remains fragmented and inconsistent, with the European Medicines Agency (EMA), the United States Food and Drug Administration (FDA), and other governing bodies struggling to keep pace with rapid AI advancements.³⁴ The lack of standardized approval pathways has led to a proliferation of AI tools with variable levels of validation, raising concerns about their safety and efficacy in real-world settings. For example, AI-based arrhythmia detection algorithms have received regulatory clearance based on retrospective datasets, yet prospective trials demonstrating clinical utility remain scarce. Furthermore, the liability of AI-driven decision-making remains legally ambiguous.

The integration of AI into actual cardiology practice requires close collaboration between clinicians, data scientists, bioethicists, and regulators to ensure tools are clinically relevant, transparent, and unbiased.³⁵ However, current AI development often occurs in industry-driven silos, prioritizing commercial viability over clinical robustness. A lack of physician involvement in AI training and validation has resulted in models that function well in theory but fail to deliver meaningful benefits in real-world practice. Additionally, explainable AI remains an unmet need—clinicians are expected to trust AI-driven recommendations without clear reasoning behind predictions, creating barriers to adoption and accountability.

Limitations, Challenges, and Future Directions

The adoption of AI in cardiology is hindered by multiple technical, clinical, and ethical challenges that limit its widespread integration into routine practice.^19,36,37 One of the most pressing issues is the lack of standardized protocols for AI implementation, leading to variability in clinical outcomes and poor reproducibility (Table 3).^25,37 Many AI models, despite demonstrating high predictive accuracy in retrospective datasets, fail to generalize when applied to diverse patient populations, partly due to limited external validation.³⁸ This problem is exacerbated in settings requiring real-time decision-making, such as emergency cardiovascular interventions, where AI-driven recommendations may lack the adaptability and contextual awareness of experienced clinicians.

Another major challenge is algorithmic bias, which arises from imbalanced training datasets that overrepresent certain demographics while underrepresenting others, leading to disparities in AI-guided clinical decisions.³⁹ Studies have shown that AI-based risk prediction models often underperform in minority populations, potentially exacerbating healthcare inequities.⁴⁰ Despite efforts to develop bias detection and correction frameworks, robust solutions remain scarce, limiting the reliability of AI-driven tools in real-world cardiology practice.

Another critical limitation is the opacity of most complex models, which function as “black boxes” with limited interpretability for clinicians, and this holds even truer when such models are protected by patents or copyright.⁴¹ The lack of explainable AI (XAI) undermines clinician trust and complicates accountability in medical decision-making. A poignant question commonly comes to mind: if an AI-driven system misguides treatment, who is responsible—the physician, the developer, or the AI system itself? Addressing these concerns requires greater model transparency, clinically interpretable outputs, and regulatory frameworks clarifying liability in AI-assisted decision-making.

The high computational costs associated with AI model development and deployment present additional barriers to implementation, particularly in resource-limited healthcare settings.³⁶ While large academic centers and high-volume hospitals may afford AI-driven clinical decision support tools, smaller institutions and underserved regions often lack the infrastructure to integrate these technologies effectively. Additionally, AI-based cardiology tools currently face fragmented regulatory oversight, with inconsistent approval pathways between the FDA, EMA, and other regulatory bodies.³⁴ The absence of standardized validation criteria has led to the premature adoption of AI-driven diagnostics and decision support systems without robust prospective trial data confirming their real-world efficacy.

Moving forward, greater emphasis must be placed on prospective validation, ensuring AI models perform reliably across diverse populations and clinical scenarios.⁴² Promising avenues for AI include its integration with robotic-assisted interventions, where automation could enhance precision in complex PCI. However, the assumption that AI-driven robotics will outperform human operators remains speculative, necessitating large-scale, independent clinical trials to assess safety and efficacy.

Similarly, AI-powered telemedicine platforms hold the promise of improving remote cardiovascular monitoring and early disease detection, particularly in heart failure management.³⁰ However, these systems must be rigorously tested to ensure they provide actionable insights without overburdening clinicians with false positives or generating unnecessary interventions. Another area of interest is the potential role of AI in regenerative cardiology, where predictive modeling could aid stem cell therapy optimization and tissue engineering strategies.

Case Studies and Real-World Implementations

Artificial Intelligence has been successfully implemented in various real-world cardiology settings, with favorable results in terms of disease prediction, diagnosis, and management.^1,27 For instance, AI-powered diagnostic tools have improved the detection of congenital heart diseases, leading to earlier interventions and better patient outcomes.⁴³

Additionally, AI has been integrated into cardiac imaging, assisting in the interpretation of echocardiograms, MRIs, and CT scans.⁴⁴ This integration has resulted in more consistent and accurate readings, aiding clinicians in making informed decisions.

Predictive analytics has been employed to forecast cardiovascular risk and guide chronic disease management, but real-world deployment has faced challenges.³³ Indeed, AI-based prediction models for cardiovascular readmissions may appear accurate in the short term or in “bread and butter” patients but prove inaccurate over time without hands-on clinician oversight and expertise when focusing on niche, but actually quite common, patient subgroups.⁴⁵

Indeed, automated imaging analysis by non-experts could enable healthcare providers with limited imaging training to leverage AI for non-invasive diagnostics. For example, AI-assisted lung ultrasound interpretation for heart failure patients allowed general practitioners to detect pulmonary congestion with 80% sensitivity, yet false positives led to unnecessary interventions, demonstrating the need for human validation before AI-generated insights are acted upon.³³

Conclusion

The increasing integration of AI in cardiovascular medicine is not yet mirrored by evidence of its favorable real-world impact. While AI-driven tools enhance diagnostics, risk prediction, and interventional procedures, their clinical reliability, generalizability, and cost-effectiveness remain questionable. Most AI models excel in controlled settings but fail in diverse real-world populations, raising concerns about bias, transparency, and accountability. Despite claims of automation and precision, AI often introduces new inefficiencies, misclassifications, and workflow disruptions. The lack of explainability remains a major barrier, as clinicians struggle to trust opaque, algorithm-driven recommendations. Moreover, regulatory inconsistencies and high implementation costs restrict accessibility to AI, disproportionately favoring well-funded institutions. Future progress hinges on rigorous external validation, clear regulatory oversight, and interdisciplinary collaboration. AI should remain a clinical adjunct, not a substitute for human expertise. Without critical evaluation and evidence-based integration, AI risks becoming another overhyped technology with marginal real-world benefit.

Acknowledgements:

This manuscript was drafted with the assistance of artificial intelligence tools, including ChatGPT 4 (OpenAI, San Francisco, CA, USA), in keeping with established best practices (Biondi-Zoccai G, editor. ChatGPT for Medical Research. Torino: Edizioni Minerva Medica; 2024). The final content, including all conclusions and opinions, has been thoroughly revised, edited, and approved by the authors. The authors take full responsibility for the integrity and accuracy of the work and retain full credit for all intellectual contributions. Compliance with ethical standards and guidelines for the use of artificial intelligence in research has been ensured.

Footnotes

Peer-review: Internally peer-reviewed.

Author Contributions: Concept – G.B.Z.; Design – G.B.Z.; Supervision – U.M., Ç.E., S.C., P.L., F.V.; Resources – G.B.Z, F.D.A., S.C., P.L., F.V.; Materials – G.B.Z, F.D.A. S.G.; Data Collection and/or Processing – G.B.Z., F.D.A. S.G.; Analysis and/or Interpretation – G.B.Z., F.D.A. S.G., U.M., Ç.E., S.C., P.L., F.V; Literature Search – G.B.Z., F.D.A. S.G.; Writing – G.B.Z., F.D.A. S.G.; Critical Review – U.M., Ç.E., S.C., P.L., F.V.

Disclosure: Giuseppe Biondi-Zoccai has consulted for Abiomed, Advanced Nanotherapies, Aleph, Amarin, Balmed, Cardionovum, Crannmedical, Endocore Lab, Eukon, Guidotti, Innovheart, Meditrial, Menarini, Microport, Opsens Medical, Terumo, and Translumina, outside the present work. Denisa Muraru reports research support and speakers’ fees from GE Healthcare and Philips Medical Systems, outside the present work. All other authors report no conflict of interest.

References

1. Lüscher TF, Wenzl FA, D’Ascenzo F, Friedman PA, Antoniades C. Artificial intelligence in cardiovascular medicine: clinical applications. Eur Heart J. 2024;45(40):4291-4304.
2. Gallone G, Bruno F, D’Ascenzo F, DE Ferrari GM. What will we ask to artificial intelligence for cardiovascular medicine in the next decade?. Minerva Cardiol Angiol. 2022;70(1):92-101.
3. Candela-Juan C, Ciraj-Bjelac O, Sans Merce M. Use of out-of-field contact shielding on patients in medical imaging: a review of current guidelines, recommendations and legislative documents. Phys Med. 2021;86():44-56.
4. D’Ascenzo F. Artificial intelligence in cardiology: the next big thing?. Minerva Cardiol Angiol. 2022;70(1):65-66.
5. D’Ascenzo F, De Filippo O, Gallone G. Machine learning-based prediction of adverse events following an acute coronary syndrome (PRAISE): a modelling study of pooled datasets. Lancet. 2021;397(10270):199-207.
6. Khera R, Oikonomou EK, Nadkarni GN. Transforming cardiovascular care with artificial intelligence: from discovery to practice: JACC state-of-the-art review. J Am Coll Cardiol. 2024;84(1):97-114.
7. Biondi-Zoccai G. . ChatGPT for Medical Research. 2024;():-.
8. Lüscher TF, Wenzl FA. The cardiologist in the age of artificial intelligence: what is left for us?. Cardiovasc Res. 2024;120(14):e57-e59.
9. Bos JM, Liu K, Attia ZI, Noseworthy PA, Friedman PA, Ackerman MJ. Deep neural network analysis of the 12-lead electrocardiogram distinguishes patients with congenital long QT syndrome from patients with acquired QT prolongation. Mayo Clin Proc. 2025;100(2):276-289.
10. Seetharam K, Balla S, Bianco C. Applications of machine learning in cardiology. Cardiol Ther. 2022;11(3):355-368.
11. Frati G, Carnevale R, Nocella C. Profiling the acute effects of modified risk products: evidence from the SUR-VAPES (Sapienza University of Rome-Vascular Assessment of Proatherosclerotic Effects of Smoking) cluster study. Curr Atheroscler Rep. 2020;22(2):8-.
12. Testa A, Biondi-Zoccai G, Anticoli S. Cluster analysis of weather and pollution features and its role in predicting acute cardiac or cerebrovascular events. Minerva Med. 2022;113(5):825-832.
13. Mohammadi T, D’Ascenzo F, Pepe M. Unsupervised machine learning with cluster analysis in patients discharged after an acute coronary syndrome: insights from a 23,270-patient study. Am J Cardiol. 2023;193():44-51.
14. Rai HM, Chatterjee K. Hybrid CNN-LSTM deep learning model and ensemble technique for automatic detection of myocardial infarction using big ECG data. Appl Intell. 2022;52(5):5366-5384.
15. Stamate E, Piraianu AI, Ciobotaru OR. Revolutionizing cardiology through artificial intelligence-big data from proactive prevention to precise diagnostics and cutting-edge Treatment-A comprehensive review of the past 5 years. Diagnostics (Basel). 2024;14(11):1103-.
16. Duchateau N, King AP. . AI and Big Data in Cardiology: A Practical Guide. 2023;():-.
17. Dhingra LS, Aminorroaya A, Sangha V. Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study. Eur Heart J. 2025;46(11):1044-1053.
18. Rajendran S, Pan W, Sabuncu MR, Chen Y, Zhou J, Wang F. Patchwork learning. arXiv. 2023;():-.
19. Asselbergs FW, Lüscher TF. Trustworthy implementation of artificial intelligence in cardiology: a roadmap of the European Society of Cardiology. Eur Heart J. 2024;46(8):677-679.
20. Hidayaturrohman QA, Hanada E. Impact of data pre-processing techniques on XGBoost model performance for predicting all-cause readmission and mortality among patients with heart failure. BioMedInformatics. 2024;4(4):2201-2212.
21. Fortuni F, Ciliberti G, De Chiara B. Advancements and applications of artificial intelligence in cardiovascular imaging: a comprehensive review. Eur Heart J Imaging Methods Pract. 2024;2(4):qyae136-.
22. . The TRANSFORM trial. . ;():-. https://reports.mountsinai.org/article/card2024-transform
23. Chandra Sekar PK, Veerabathiran R. Emerging technologies and applications in multimodality imaging for ischemic heart disease: current state and future of artificial intelligence. Explor Cardiol. 2024;2(6):253-264.
24. Nudi F, Schillaci O, Biondi-Zoccai G, Iskandrian AE. . Hybrid Cardiac Imaging for Clinical Decision-Making From Diagnosis to Prognosis. 2023;():-.
25. van Smeden M, Heinze G, Van Calster B. Critical appraisal of artificial intelligence-based prediction models for cardiovascular disease. Eur Heart J. 2022;43(31):2921-2930.
26. Prabhakaran SP. Cloud-enabled ai infrastructure in healthcare: a systematic review of clinical decision support and workflow optimization. IJRCAIT. 2024;7(2):2430-2441.
27. Beyar R, Davies JE, Cook C, Dudek D, Cummins PA, Bruining N. Robotics, imaging, and artificial intelligence in the catheterisation laboratory. EuroIntervention. 2021;17(7):537-549.
28. Sazzad F, Ler AAL, Furqan MS. Harnessing the power of artificial intelligence in predicting all-cause mortality in transcatheter aortic valve replacement: a systematic review and meta-analysis. Front Cardiovasc Med. 2024;11():1343210-.
29. Wagener M, Onuma Y, Sharif R, Coen E, Wijns W, Sharif F. Features and limitations of robotically assisted percutaneous coronary intervention (R-PCI): a systematic review of R-PCI. J Clin Med. 2024;13(18):5537-.
30. Tolu-Akinnawo O, Ezekwueme F, Awoyemi T. Telemedicine in cardiology: enhancing access to care and improving patient outcomes. Cureus. 2024;16(6):e62852-.
31. Khanam M, Akther S, Mizan I. The potential of artificial intelligence in unveiling healthcare’s future. Cureus. 2024;16(10):e71625-.
32. Pepe M, Larosa C, Rosa I. Degenerative severe aortic stenosis and concomitant coronary artery disease: what is changing in the era of the “transcatheter revolution”?. Curr Atheroscler Rep. 2020;22(5):17-.
33. Baloescu C, Bailitz J, Cheema B. Artificial intelligence-guided lung ultrasound by nonexperts. JAMA Cardiol. 2025;10(3):245-253.
34. Wellnhofer E. Real-World and regulatory perspectives of artificial intelligence in cardiovascular imaging. Front Cardiovasc Med. 2022;9():890809-.
35. Gruson D, Bernardini S, Dabla PK, Gouget B, Stankovic S. Collaborative AI and laboratory medicine integration in precision cardiovascular medicine. Clin Chim Acta. 2020;509():67-71.
36. Cau R, Pisu F, Suri JS, Saba L. Addressing hidden risks: systematic review of artificial intelligence biases across racial and ethnic groups in cardiovascular diseases. Eur J Radiol. 2025;183():111867-.
37. Al-Kfairy M, Mustafa D, Kshetri N, Insiew M, Alfandi O. Ethical challenges and solutions of generative AI: an interdisciplinary perspective. Informatics. 2024;11(3):58-.
38. Santoro F, Núñez-Gil IJ, Viana-Llamas MC. Anticoagulation therapy in patients with coronavirus disease 2019: results from a multicenter international prospective registry (health outcome predictive evaluation for corona virus disease 2019 [HOPE-COVID19]). Crit Care Med. 2021;49(6):e624-e633.
39. Liu S, Russo C, Suero Molina E, Di Ieva A. Artificial intelligence methods. Adv Exp Med Biol. 2024;1462():21-38.
40. Nolin-Lapalme A, Corbin D, Tastet O, Avram R, Hussin JG. Advancing fairness in cardiac care: strategies for mitigating bias in artificial intelligence models within cardiology. Can J Cardiol. 2024;40(10):1907-1921.
41. Abgrall G, Holder AL, Chelly Dagdia Z, Zeitouni K, Monnet X. Should AI models be explainable to clinicians?. Crit Care. 2024;28(1):301-.
42. Gifari MW, Machino-Ohtsuka T, Machino T, Hassan M, Suzuki K. On the feasibility of a robotic probe manipulator for echocardiography in the prone position. Front Robot AI. 2024;11():1474077-.
43. Reddy CD, Van den Eynde J, Kutty S. Artificial intelligence in perinatal diagnosis and management of congenital heart disease. Semin Perinatol. 2022;46(4):151588-.
44. Massalha S, Clarkin O, Thornhill R, Wells G, Chow BJW. Decision support tools, systems, and artificial intelligence in cardiac imaging. Can J Cardiol. 2018;34(7):827-838.
45. Van Grootven B, Jepma P, Rijpkema C. Prediction models for hospital readmissions in patients with heart disease: a systematic review and meta-analysis. BMJ Open. 2021;11(8):e047576-.