New Insights into Valvular Interventions in the Era of Artificial Intelligence

On behalf of the editorial board, I am pleased and proud to introduce the sixth issue in the life of the Journal of the Heart Valve Society (JHVS), which is actually the second issue of the Volume 3 published in 2026.

This issue primarily focuses on new insights into valve interventions in the era of artificial intelligence (AI). Indeed, we are living in an era of AI and computational modeling / simulation, and the role and contribution of these tools will rapidly expand in the near future and will have major implications in research and in the clinical management of valvular heart diseases (VHDs). We previously published in the journal several articles dealing with this topic,^1–3 but today, we present a full issue that is dedicated, in large part, to the role of AI and computational modeling in valve interventions. I do not think AI will replace the clinicians or scientists in this field, but it will certainly enhance and complement them. Ginni Rometty, former Chief Executive Officer at IBM stated “Some people call this artificial intelligence, but the reality is this technology will enhance us. So instead of artificial intelligence, I think we’ll augment our intelligence.” Indeed, AI and mathematical modeling will simplify, accelerate, and enhance determination of VHD severity and risk stratification and will help to individualize and optimize the timing and type of intervention. Also, and importantly, AI will help to reduce disparity and inequity in access to care for patients with VHD. ⁴

The articles published in this issue of the JHVS provide a compelling overview of the great potential of AI in the context of valve interventions. The current issue of the JHVS, which presents: 4 original articles, 2 review articles, 1 viewpoint, 1 technical note, and 1 special editorial, contributes to advance the knowledge on the application of AI and mathematical modeling to enhance risk stratification and outcomes of valve interventions. And given the topic of this issue, I have asked Chat GPT to generate a figure to summarize the content of this issue (Figure 1).

OPEN IN VIEWER

Figure 1.

JHVS issue on AI and computational modeling across the continuum of valve care.

In a compelling original article, Yueqing Sun, Zahra Keshavarz-Motamed, and colleagues from McMaster University, Hamilton, Ontario, Canada and Hospital Universitario Marqués de Valdecilla Servicio de Medicina Interna, Valdecilla, Spain, uncover novel predictors of transcatheter aortic valve replacement (TAVR) futility and mid-term outcomes using interpretable machine learning. Traditional risk models are poorly predictive of TAVR futility, failing to identify high-risk patients that will not benefit from the procedure. The authors proposed an interpretable machine learning model, which identifies a panel of novel predictors for poor mid-term TAVR outcomes, significantly outperforming traditional risk scores. In particular, frailty and aortic stiffness appeared to be very strong predictors of TAVR futility. Interestingly, the authors also observed a “cholesterol paradox”, where in this high-risk population, lower LDL-cholesterol was strongly predictive of poor outcomes.

This article is accompanied by an elegant editorial by Théo Pezel and Solenn Toupin from MIARCL.ai laboratory, University Hospital of Lariboisiere, Paris, France. The editorialists underline that this study represents an important early step toward and sets the stage for future work aimed at refining, validating, and ultimately operationalizing these insights to improve patient care. The underlying vision, a more personalized and frailty-informed approach to TAVR risk assessment, is both timely and compelling.

In an original research article, Mosarrof Hossen, Huseyin Cagatay Yalcin, Abdulrahman Alnabti, and colleagues from Qatar University, Qatar developed a deep-learning model for the prediction of patient-specific aortic wall mechanical stress prediction following TAVR. The authors implemented Graph Neural Network models for predicting patient-specific stress outcomes, specifically contact pressure and Von Mises stress, following TAVR.

This article is accompanied by a very interesting editorial from Imran Shah, Alessandro Veneziani, and Lakshmi P. Dasi, from Georgia Institute of Technologies and Emory University, Atlanta, USA. In this editorial, the authors state that the convergence of high-fidelity computational biomechanics, physics-driven model reduction, geometric deep learning, and large-scale clinical imaging datasets is creating the conditions for a transformation in how clinical teams can plan for and evaluate structural heart interventions. The article by Hossen et al serves as a great demonstration of this evolution driven by AI and computational modeling.

In a very interesting original article, Taylor Becker, Lakshmi Prasad Dasi, and colleagues from Ohio State University College of Medicine, Columbus, OH, Baylor Scott & White The Heart Hospital, Plano, TX, Marcus Valve Center, Piedmont Heart Institute, Georgia Institute of Technology, and Emory University, Atlanta, GA, demonstrate that blinded 3D computational modeling of peak areal stretch predicts patient-specific risk of aortic root rupture during TAVR and outperforms traditional planning metrics alone. Despite its lethality, aortic root rupture during TAVR remains difficult to predict using standard pre-procedural predictive models. This study demonstrates that computationally derived peak areal stretch identifies patients at risk with high accuracy. These findings highlight critical limitations in current planning approaches and offer a data-driven method to overcome these limitations.

This article is accompanied by an editorial by Lyes Kadem from Concordia University, Montréal, and Viktoria Stanova from Institut Universitaire de Cardiologie et de Pneumologie de Québec, Université Laval, Québec, Canada, who concluded that Becker et al report a methodologically rigorous approach to a clinical problem that existing pre-procedural metrics still do not fully capture and that peak areal stretch represents a genuine step forward in pre-procedural risk stratification for TAVR.

In an original research article, John Carney, Koray Potel, and colleagues from Minneapolis University, Twin Cities, USA present a hybrid surgical approach for transcatheter mitral valve replacement (TMVR) in the domestic sheep model. Sheep are the primary in vivo model for testing novel heart valve devices, though their mitral anatomy complicates TMVR device deployment. The authors developed a hybrid surgical approach for TMVR in sheep to enable long-term evaluation prior to clinical investigation. This novel approach may overcome the model-related morbidity and mortality associated with performing TMVR percutaneously in the sheep model, and facilitate long-term device evaluation required for regulatory approval.

This article is accompanied by an insightful editorial by Guillaume Guimbretière and Romain Capoulade from Institut du Thorax – INSERM, Nantes, France, who considered that the hybrid surgical approach proposed by Carney et al provides a valuable and reproducible framework for long-term evaluation of transcatheter mitral valves in large animal models. They stated that, as TMVR technologies continue to evolve, such methodological advances will be essential to bridge the gap between innovation and safe clinical translation.

In this issue, we publish a review article by Jairo Sánchez-Blanco and Luz Clemencia Zárate-Correa and colleagues from Icesi University, Cali, Colombia, and Laval University, Québec, Canada, who present an evidence-based update on early intervention versus clinical surveillance in asymptomatic patients with severe aortic stenosis (AS). The authors conclude that contemporary evidence supports a paradigm shift toward earlier intervention in selected asymptomatic patients with severe AS, guided by multimodal assessment of myocardial structural and functional consequences. However, the identification of optimal intervention thresholds remains an active area of clinical investigation, requiring careful integration of hemodynamic markers, advanced imaging, and individual risk profiles.

In a comprehensive review article by Darshan Reddy and Robin Kinsley from Lenmed Ethekwini Hospital and Heart Centre, University of KwaZulu-Natal Nelson R Mandela School of Medicine, Durban, South Africa presents a review of the pathogenesis in juvenile rheumatic mitral valve disease with implications for valve repair. This review underlines the importance of immune-mediated connective tissue damage in the development of rheumatic heart valve disease. The authors conclude that juvenile rheumatic mitral valve disease should be considered a separate entity from the chronic or burnt-out valvulitis seen in older patients, with distinctive morphological and echocardiographic features.

In this issue of JHVS, Patrizio Lancellotti and colleagues from the EuroValve Consortium present an insightful viewpoint on Heart Valve Centers in the era of data-driven and AI-enabled valve care. The key learning points of this article are the following: (i) Contemporary valve care requires structured organization integrating Heart Valve Clinics, multidisciplinary Heart Teams, and institutional Centers of Reference within coordinated networks. (ii) AI enhances imaging quantification, risk stratification, and longitudinal trajectory assessment but must remain embedded within expert clinical governance. (iii) Network-based, expert-led valve care improves consistency, equity, and sustainability across the lifetime management of VHD. This is a very interesting viewpoint by an international panel of experts, which I strongly encourage you to read.

In a very well-illustrated technical note, Amine Fikani and colleagues from Medical Center Hotel-Dieu de France Hospital, Saint Joseph University of Beirut, Beirut, Lebanon, present a new surgical technique that offers a simple and effective bailout option for mitral valve replacement in calcified mitral valve and annulus, which is often very challenging technically.

Finally, and sadly, we are publishing, in this issue, an In Memoriam for Robert A. Levine (1953-2026) written with heart and soul by Adrien Lupieri, Victor Nizet, Walderez Dutra, Maria Carmo Nunes, Elana Aikawa for the Leducq PRIMA network. Dr Levine was one of the most brilliant and productive investigators in the field of VHD. He was also a visionary and a leader in research and innovation, a great mentor, and a true gentleman. His endless curiosity and new ideas of research, his energy and enthusiasm have been a great source of inspiration and motivation for the VHD research and clinical care community. He will be deeply missed.

I hope that you will enjoy and learn a lot from this very rich issue of the journal that we are publishing today on such an important and timely topic: ie, the role of AI and computational modeling in valve interventions. Given that we receive a large number of articles on this topic, we plan to publish another focused issue in the near future. However, the next priority topic that we are considering for the last focus issue of this year is congenital heart valve disease. So, we strongly encourage you and your colleagues to submit articles related to this topic.