Integrating Artificial Intelligence for Mitral Regurgitation Assessment: A systematic review and meta-analysis

Integrating Artificial Intelligence for Mitral Regurgitation Assessment: A systematic review and meta-analysis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Integrating Artificial Intelligence for Mitral Regurgitation Assessment: A systematic review and meta-analysis Lies Dina Liastuti, Nathaniel Gilbert Dyson, Haifa Mayang Lestari, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6626740/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Mitral regurgitation (MR) is a prevalent and potentially progressive cardiovascular condition, necessitating early detection to facilitate timely intervention and optimize patient outcomes. The increasing demand for efficient and precise diagnostic strategies has underscored the potential of artificial intelligence (AI) in clinical practice. By leveraging advanced AI algorithms, automated MR screening has the capacity to enhance the detection and classification of disease severity, thereby assisting clinicians in making well-informed decisions. This study aims to evaluate the accuracy and efficacy of AI-based echocardiographic analysis in the early diagnosis of MR. Main Text A comprehensive literature search was conducted across five databases, PubMed, Scopus, ScienceDirect, ProQuest, and Cochrane. Studies employing AI algorithms to analyze echocardiographic images for MR detection and severity classification were included. The methodological quality of each study was assessed using the QUADAS-2 tool for diagnostic accuracy studies. A quantitative meta-analysis was performed utilizing Meta-DiSc with a random-effects model. In total, nine studies met the inclusion criteria. Utilization of AI in echocardiographic detection of MR yield a pooled sensitivity of 0.85 (95% CI: 0.86–0.86), specificity of 0.83 (95% CI: 0.82–0.83), and an area under the curve (AUC) of 0.9745. Accurate detection and severity classification of MR could significantly improve the efficiency of treatment strategies. By reducing reliance on specialized personnel and enabling the use of portable imaging devices, AI can lower operational costs and expand access to high-quality diagnostics. Furthermore, AI integration has the potential to streamline clinical workflows, decrease diagnostic delays, and optimize resource allocation. However, successful implementation requires addressing challenges related to model generalizability, regulatory standards, clinician training, and integration into existing healthcare systems. Conclusion In conclusion, AI-assisted echocardiographic analysis presents a promising advancement in MR diagnostics, with the potential to enhance healthcare accessibility, particularly in resource-limited settings. artificial intelligence deep learning echocardiography mitral regurgitation screening Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Mitral regurgitation (MR) is one of the most prevalent valvular heart diseases worldwide and represents a significant contributor to cardiovascular morbidity and mortality. Recent data suggest that the global prevalence of moderate-to-severe MR is approximately 0.67% (95% CI, 0.33–1.11), based on population-based studies including more than 6 million individuals. 1 The clinical impact of MR is substantial, as it often leads to symptoms such as exertional dyspnea and fatigue, particularly in advanced stages, thereby markedly diminishing patients’ quality of life. 2 In regions where RHD remains endemic, MR is a common sequela, often affecting younger individuals and contributing substantially to early morbidity and mortality. 3 MR has serious financial repercussions for healthcare systems in addition to its clinical effects. According to recent data from Maryland, USA, individuals undergoing medical therapy for MR had to pay an average of $ 23,575 a year. These patients spent 3.1 more days in the hospital annually and had higher annual healthcare expenses ( $ 10,559) than those without MR. The average expenses for patients having transcatheter or surgical mitral valve replacement or repair were $ 63,108 and $ 47,943 for the treatment year, respectively. In contrast to continuing medical care, yearly costs for both groups were much lower in the second and third years after these treatments. 4 Since the underlying cause has a significant influence on both treatment choices and long-term results, determining the MR etiology accurately is essential for directing suitable management techniques. Infectious endocarditis, rheumatic heart disease, coronary artery disease, MVP, flail leaflets, drug-induced valvulopathy, and connective tissue diseases are among the intrinsic defects of the mitral valve apparatus that usually cause primary MR. While chronic MR frequently develops gradually over time, acute severe MR can result from abrupt structural breakdown, such as chordae tendineae rupture or papillary muscle dysfunction. 5 According to current clinical guidelines, patients with suspected MR should have transthoracic echocardiography for diagnosis and then routine echocardiographic monitoring at the right intervals to avoid irreversible damage to the pulmonary circulation and ventricles, which can happen without cause. 6 MR can be diagnosed noninvasively using auscultation or patient interview, however neither technique is objective, and examiner accuracy varies greatly. Even though echocardiography is the gold standard, doing it takes a lot of time and skill. 7 Therefore, more objective techniques to assess echocardiographic imaging of MR patients are desperately needed. Despite the recognized importance of early diagnosis, the implementation of widespread echocardiographic screening programs in low- and middle-income countries (LMICs) remains a significant challenge. These regions, which bear a disproportionate burden of rheumatic heart disease, often lack adequate imaging infrastructure, specialized personnel, and standardized protocols. 8 – 10 Various strategies have been proposed to address these barriers, including the deployment of portable ultrasound devices, simplification of imaging workflows, and the training of non-physician healthcare providers. Concurrently, rapid advancements in artificial intelligence (AI) have created new opportunities to enhance cardiac imaging. AI-powered systems are being developed to automate both image acquisition and interpretation, potentially reducing operator dependency and increasing diagnostic consistency across diverse clinical settings. 11 These innovations are particularly relevant in resource-limited environments, where access to trained echocardiographers is restricted. Given the increasing global burden of MR and the limitations of conventional diagnostic approaches, AI-assisted tools may offer a scalable and efficient alternative for early detection and risk stratification. This review aims to answer whether AI-based echocardiographic assessment can match or surpass traditional methods in accuracy and reliability, thereby supporting wider adoption in diverse healthcare settings. By synthesizing current evidence, this review seeks to inform clinical guidelines and support the equitable integration of AI technologies into cardiovascular care. Methods Information sources and search strategy This systematic review and meta-analysis was conducted following a pre-registered protocol with the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42024552951). To find pertinent papers published up to April 27, 2025, a thorough literature search was conducted across five major electronic databases: PubMed, Cochrane Library, ProQuest, Scopus, and ScienceDirect. In addition to free-text terms like "Mitral Valve Insufficiency," "Artificial Intelligence," and "Machine Learning," the search strategy combined Medical Subject Headings (MeSH) with echocardiography-specific terms like "Echocardiography, Three-Dimensional" and "Echocardiography, Transesophageal." This strategy was created to guarantee that all relevant material was included and to optimize the search's sensitivity and specificity. Duplicate entries were found and eliminated once all recovered records were loaded into the EndNote X9 reference management program. To determine study eligibility, two reviewers (NG and HM) independently evaluated abstracts and titles. The final inclusion of articles that were judged possibly relevant was decided by full-text examination. A third reviewer was consulted or spoken with in order to settle any disputes that arose throughout the screening or selection process. At the full-text stage, the reasons for exclusion were also well recorded. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy's methodological requirements were followed in the conduct of this review. 12 Table 1 Table of keywords for literature search Database Search strategy Hits Pubmed ("Mitral Valve Insufficiency"[Mesh] OR "Mitral Valve Insufficiency/diagnosis"[Mesh] OR "Mitral Valve Insufficiency/diagnostic imaging"[Mesh]) AND (((("Artificial Intelligence"[Mesh]) OR "Machine Learning"[Mesh]) OR "Supervised Machine Learning"[Mesh]) OR "Support Vector Machine"[Mesh]) OR "Unsupervised Machine Learning"[Mesh] AND ("Echocardiography"[Mesh] OR "Echocardiography, Stress"[Mesh] OR "Echocardiography, Four-Dimensional"[Mesh] OR "Echocardiography, Three-Dimensional"[Mesh] OR "Echocardiography, Doppler, Pulsed"[Mesh] OR "Echocardiography, Doppler, Color"[Mesh] OR "Echocardiography, Transesophageal"[Mesh] OR "Echocardiography, Doppler"[Mesh]) 57 results Cochrane (mitral valve regurgitation OR mitral valve insufficiency OR mitral regurgitation in Title Abstract Keyword) AND (artificial intelligence OR machine learning OR supervised machine learning OR unsupervised machine learning OR deep learning in Title Abstract Keywords) AND (echocardiography OR echocardiogram in Title Abstract Keywords) 2 results Proquest (Mitral Valve Insufficiency OR Mitral Regurgitation OR Mitral Valve Regurgitation) AND (Artificial Intelligence OR Machine Learning OR Supervised Machine Learning OR Support Vector Machine OR Unsupervised Machine Learning) AND (echocardiography) AND (accuracy OR area under the curve OR AUC OR sensitivity OR specificity) AND (heart function OR ventricular function) 1,284 results Scopus TITLE-ABS-KEY ( ( mitral AND valve AND insufficiency OR mitral AND regurgitation OR mitral AND valve AND regurgitation ) AND ( artificial AND intelligence OR machine AND learning OR supervised AND machine AND learning OR support AND vector AND machine OR unsupervised AND machine AND learning ) AND ( echocardiography ) ) 10 results Science Direct ("Mitral Valve Insufficiency" OR "Mitral Regurgitation" OR "Mitral Valve Regurgitation") AND ("Artificial Intelligence" OR "Machine Learning" OR "Supervised Machine Learning" OR "Support Vector Machine" OR Unsupervised Machine Learning) AND "echocardiography" 265 results Study eligibility criteria This review included observational studies—specifically cross-sectional and cohort designs—that investigated the diagnostic accuracy of AI or machine learning algorithms in identifying MR. Eligible studies were required to utilize a recognized reference standard, such as interpretation from echocardiographers or cardiologists for MR diagnosis. Inclusion was limited to articles published in English. There were no restrictions regarding patient age at diagnosis or the specific AI methodology applied. Studies that did not clearly define the reference standard were excluded. For studies that assessed multiple types of valvular heart disease, only data pertaining to MR were extracted and analyzed. Data extraction Two independent reviewers (NG and HM) performed data extraction using a standardized, pretested form to ensure methodological rigor and consistency. Extracted information included bibliographic details (author names, publication year, and study title), geographical location, study design, clinical setting, and sample size—distinguishing between training and testing datasets where applicable. Patient selection criteria were also recorded. In addition, technical characteristics of the AI/ML models were documented, including model architecture, input data type, reference standard used for model validation (e.g., expert echocardiographers or cardiologists), diagnostic performance indicators (e.g., sensitivity, specificity, area under the curve [AUC]), and details of internal or external validation procedures. Any disagreements regarding study inclusion, data abstraction, or interpretation were resolved through iterative discussion. If consensus could not be achieved, a third reviewer (LD) was consulted to provide adjudication. This triangulated review process ensured methodological transparency and reduced the risk of bias. Quality assessment The Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) instrument was used by two reviewers (NG and HM) to independently evaluate the risk of bias. 13 Four areas are assessed by this framework: flow and timing, reference standard, index test, and patient selection. Concerns about applicability were also evaluated for the first three areas. Any item inside each domain that was deemed to be high risk resulted in the domain as a whole being classified as high risk.Consensus talks were used to settle disagreements on risk of bias assessments. A third reviewer (LD) acted as an arbitrator to complete evaluations when needed. A thorough and objective evaluation of the study's quality was guaranteed by this methodical and planned methodology. Statistical analysis Quantitative synthesis of diagnostic performance was performed using Meta-DiSc version 1.4. For each included study, pooled estimates of sensitivity and specificity were calculated, along with their corresponding 95% confidence intervals (CIs). Summary receiver operating characteristic (SROC) curves were generated to provide an overall measure of diagnostic accuracy, and results were visually presented using forest plots and SROC curves. To investigate potential sources of heterogeneity, subgroup analyses were conducted, with particular attention to the types of artificial intelligence (AI) or machine learning (ML) algorithms utilized. Where appropriate, sensitivity analyses were carried out to evaluate the robustness of the pooled estimates and to determine the impact of individual studies on the overall results. Results Study characteristics The initial search yielded 1,618 articles, of which 1,555 were deemed eligible for screening after removing 63 duplicates. Following title and abstract screening, 1,278 studies were excluded due to irrelevant title. Of the remaining 277 studies, 246 were further excluded due to irrelevant abstract screening, leaving 31 studies for full-text review. After detailed evaluation, 20 studies were excluded: 7 were not original research, and 15 lacked accessible full-text versions. Finally, 9 studies met the inclusion criteria and were included in the review (see Fig. 1 for the study selection flowchart). The characteristics of the nine studies included in this review are summarized in Table 2 . All studies employed were observational studies, with seven adopting a cross-sectional approach and two utilizing retrospective cohort methodologies. Cross-sectional studies assessed the diagnostic performance of AI algorithms at a single time point, while retrospective cohort studies enabled evaluation of algorithm performance over previously collected datasets. Specifically, five studies were performed in high-income countries (four in the USA and one in Hong Kong), while three were conducted in middle-income countries (two in China and one in Iran). The included studies analyzed echocardiographic images from diverse datasets. Various AI methodologies were employed, including convolutional neural networks (CNN), deep learning (DL), support vector machines (SVM), machine learning (ML), and artificial intelligence-based ultrasound systems (AIUS). We further categorized the included studies based on their primary objective, distinguishing between those that evaluated the diagnostic accuracy of AI algorithms for MR detection and those that focused on grading the severity of mitral regurgitation. Bias assessment The risk of bias assessment for the included studies is summarized in Fig. 2 . Overall, the majority of studies demonstrated a low risk of bias across all evaluated domains. Most studies were rated as low risk for patient selection, index test, reference standard, and flow and timing. However, some concerns were noted in specific domains for a few studies. Zhang et al. (2021) and Moghaddasi et al. (2016) showed some concerns regarding the reference standard, while Brown et al. (2024), Long et al. (2024), Zhang et al. (2021), and Moghaddasi et al. (2016) had some concerns in the flow and timing domain. Despite these minor concerns, no issues were identified regarding the applicability of the index test or patient selection. These findings suggest that the included studies are methodologically sound and that their results are relevant and appropriately aligned with the objectives of this review. AI algorithms and reference standards A variety of artificial intelligence (AI) algorithms were utilized across the included studies, reflecting the evolving landscape of machine learning applications in cardiovascular imaging. Convolutional neural networks (CNNs) and deep learning (DL) models were the most frequently employed, particularly for tasks involving MR detection and image-based classification. For example, Brown et al. (2024) and Edwards et al. (2023) implemented CNNs for automated MR detection, while Long et al. (2024) and Vrudhula et al. (2024) used broader DL approaches with large-scale echocardiographic datasets. Support vector machines (SVMs) were also applied, notably in studies by Yang et al. (2022) and Moghaddasi (2016), primarily for MR severity grading and classification tasks. Additionally, other machine learning (ML) techniques and specialized algorithms such as artificial intelligence-based ultrasound systems (AIUS) were reported, as seen in Jin et al. (2016) and Sadeghpour et al. (2025). The reference standards for evaluating AI performance varied among studies, reflecting differences in clinical practice and available expertise. Most studies used diagnoses established by experienced cardiologists or echocardiographers as the reference standard, either individually or through expert consensus panels. For instance, Brown et al. (2024) and Jin et al. (2016) validated their AI models against the combined assessments of echocardiographers and cardiologists, while others, such as Long et al. (2024) and Vrudhula et al. (2024), relied on cardiologist adjudication alone. In some cases, expert echocardiographers provided detailed grading of MR severity, as in Zhang et al. (2021) and Moghaddasi (2016). Table 2 Summary of study characteristics Study Study Design Country Income region Training dataset Testing dataset AI Algorithm MR Detection Brown et al 11 , 2024 Cross-sectional USA High 95 416 CNN Long et al 14 , 2024 Cross-sectional USA High 43,811 17,878 DL Vrudhula et al 15 , 2024 Cross-sectional USA High 80,833 46,890 DL Edwards et al 16 , 2023 Cross-sectional USA High 66,330 11,730 CNN Yang et al 17 , 2022 Retrospective cohort China Middle 777 151 SVM Jin et al 18 , 2016 Retrospective cohort Hong Kong High 90 90 AIUS MR Severity Grading Sadeghpour et al 19 , 2025 Cross-sectional USA High 438 438 ML Zhang et al 20 , 2021 Cross-sectional China Middle 1,132 295 CNN Moghaddasi et al 21 , 2016 Cross-sectional Iran Middle 102 102 SVM Table 3 Summary of study results Study Index test Reference standard Study outcomes Sensitivity Specificity Accuracy F1 score AUC MR Detection Brown et al 11 , 2024 CNN Echocardiographers and cardiologists N/A N/A 0.99 0.87 0.93 Long et al 14 , 2024 DL Cardiologists N/A N/A 0.82 0.8 0.98 Vrudhula et al 15 , 2024 DL Cardiologists 0.975 0.999 N/A N/A 0.916 Edwards et al 16 , 2023 CNN Cardiologists N/A N/A 0.86 0.97 0.91 Yang et al 17 , 2022 SVM Cardiologists 0.94 0.94 0.94 N/A 0.97 Jin et al 18 , 2016 AIUS Echocardiographers and cardiologists 0.90 0.97 0.89 N/A N/A MR Severity Grading Sadeghpour et al 19 , 2025 ML Cardiologists 0.96 0.98 Mild: 0.80 Moderate and severe: 0.97 N/A N/A Zhang et al 20 , 2021 CNN Echocardiographers N/A N/A I: 0.90 II: 0.87 III: 0.81 IV: 0.91 I: 0.94 II: 0.88 III: 0.85 IV: 0.89 N/A Moghaddasi et al 21 , 2016 SVM Echocardiographers Mild: 0.99 Moderate: 0.98 Severe: 0.99 Mild: 0.99 Moderate: 0.99 Severe: 0.99 N/A N/A N/A AI accuracy in MR detection Our meta-analysis demonstrated that AI-assisted echocardiographic analysis for the detection of mitral regurgitation (MR) yielded a pooled sensitivity of 0.86 (95% confidence interval [CI]: 0.86 to 0.86). This finding indicates that AI algorithms are highly effective in identifying MR across the included studies. However, substantial heterogeneity was observed, as reflected by an I² value of 99.0%, far exceeding the conventional threshold of 75% for high heterogeneity. The chi-square test further confirmed this variability (χ² = 505.33, df = 5, p < 0.0001). Despite the robust pooled sensitivity, the marked heterogeneity suggests that factors such as algorithm architecture, population characteristics, and imaging equipment may influence diagnostic performance. These results highlight the need for standardized evaluation protocols and further research to optimize and generalize the application of AI in echocardiographic MR detection. Our meta-analysis showed that AI-assisted echocardiography for the detection of mitral regurgitation (MR) achieved a pooled specificity of 0.83 (95% confidence interval [CI]: 0.82 to 0.83). This suggests that AI algorithms demonstrate good accuracy in correctly identifying individuals without MR. However, substantial heterogeneity was observed among the included studies, as indicated by an I² value of 99.4%, which is well above the conventional threshold for high heterogeneity. The chi-square test further supported this finding (χ² = 770.74, df = 5, p < 0.0001), highlighting significant variability in specificity estimates across studies. Our meta-analysis demonstrates that AI-assisted echocardiography for the detection of mitral regurgitation (MR) achieves excellent diagnostic accuracy, as reflected by an area under the curve (AUC) of 0.9745 on the symmetric summary receiver operating characteristic (SROC) curve. The AUC is a key metric for evaluating the overall performance of diagnostic tests, with values approaching 1.0 indicating outstanding discriminative ability. The observed AUC underscores the high reliability of AI algorithms in distinguishing between patients with and without MR. The precision of this estimate is supported by a standard error (SE) of 0.0181 for the AUC. Additionally, the Q* index—a summary measure of diagnostic effectiveness—was 0.9274 (SE = 0.0311), further confirming the robustness of the AI-guided approach. Collectively, these findings highlight the strong potential of AI-driven tools to enhance MR detection and support more efficient and accurate diagnostic workflows in clinical practice. AI accuracy in MR severity grading The included studies demonstrate that artificial intelligence (AI) algorithms can achieve high accuracy in grading the severity of mitral regurgitation (MR) using echocardiographic data. However, due to lack of data, we could not perform meta-analysis on the MR severity grading parameter. Three studies—Sadeghpour et al 19 , Zhang et al 20 , and Moghaddasi et al 21 —specifically evaluated the diagnostic performance of various AI models for MR severity classification. The findings from each study are displayed in Table 3 . Sadeghpour et al 19 employed a machine learning (ML) approach validated by cardiologists. Their model achieved a sensitivity of 0.96 and a specificity of 0.98 for MR severity grading. The accuracy was 0.80 for mild MR and 0.97 for moderate and severe MR categories, indicating strong performance across different severity levels. Zhang et al 20 developed a Mask R-CNN deep learning algorithm for automatic MR severity assessment using color Doppler echocardiography images, following the 2017 American Society of Echocardiography (ASE) guidelines. The study included a large, multi-center dataset and validated the model both internally and externally. The algorithm achieved classification accuracies of 0.90 for grade I (mild), 0.87 for grade II (moderate), 0.81 for grade III (moderate), and 0.91 for grade IV (severe). The corresponding specificities were 0.94, 0.88, 0.85, and 0.89, respectively. The Macro F1 and Micro F1 scores for grading were both 0.89, reflecting balanced performance across all MR grades. The study also highlighted that the AI model substantially reduced the time required for MR severity assessment compared to manual guideline-based evaluation, and maintained robust performance across different hospitals and ultrasound equipment. Moghaddasi et al 21 utilized a support vector machine (SVM) model to classify MR severity into mild, moderate, and severe categories, based on echocardiographic images. The model achieved high sensitivity and specificity for all classes: 0.99 for mild, 0.98 for moderate, and 0.99 for severe MR, with corresponding specificities of 0.99, 0.99, and 0.99, and classification accuracies exceeding 99% for each category. Discussion To the best of our knowledge, this is the first meta-analysis to systematically evaluate the diagnostic accuracy of AI models for both detection and classification of MR. Mitral regurgitation is a prevalent valvular disorder characterized by the improper closure of the mitral valve, leading to the backward flow of blood into the left atrium. 22 In order to prevent consequences like atrial fibrillation and heart failure, early and accurate diagnosis is essential. However, prompt intervention is hampered by obstacles including limited access to echocardiographic facilities and expert cardiologists, especially in under-resourced areas. By allowing non-physician healthcare practitioners to do tests, AI-driven echocardiography holds promise for closing these disparities. AI combined with portable echocardiography equipment provides a scalable and economical diagnostic method, especially in settings with limited resources. 23 , 24 Our findings underscore the significant potential of AI to improve MR diagnosis, especially in environments with limited access to specialized healthcare professionals. Echocardiographic imaging emerged as the predominant data source for AI algorithms, with convolutional neural networks and support vector machines being the most commonly utilized analytical methods. Pooled results demonstrated that AI-based echocardiography achieved high sensitivity (86%) and high specificity (83%) for MR detection. The results of this meta-analysis are consistent with findings from several other recent meta-analyses evaluating the application of artificial intelligence in echocardiographic analysis. For instance, a systematic review and meta-analysis by Liastuti et al assessed the diagnostic accuracy of AI models in detecting congenital heart disease during second-trimester fetal cardiac screening, demonstrating that AI can achieve high sensitivity and specificity in this context. 25 Similarly, another meta-analysis examined the performance of machine learning models in identifying congenital heart disease, further supporting the reliability of AI-based approaches in cardiac imaging. 26 Convolutional neural networks (CNNs), a type of deep learning model, are particularly well-suited for interpreting complicated visual data, such as echocardiograms. These algorithms can automatically recognize complex patterns in the pictures, including aberrant mitral valve motion or regurgitant blood flow, that might point to the existence of MR. Usually, the procedure starts with the gathering of video clips or echocardiograms, which are subsequently preprocessed to improve picture quality and standardize the data. Normalization of pixel values, noise reduction, and segmentation of pertinent cardiac structures, such as the left atrium and mitral valve, are examples of preprocessing procedures. This guarantees that reliable and superior input data for analysis is provided to the deep learning model. 27 , 28 Unlike traditional machine learning approaches that rely on manual feature selection, deep learning models automatically extract relevant features from the raw image data. Through multiple layers of convolutional filters, CNNs can identify subtle patterns and features that are characteristic of MR. The models are trained on large datasets of labeled echocardiography images, where each image is annotated by expert cardiologists to indicate the presence or absence of MR. During training, the model learns to associate specific visual features with MR, gradually improving its diagnostic accuracy. 29 Once trained, the deep learning model can analyze new echocardiography images and predict whether MR is present. Some advanced models are also capable of grading the severity of MR, classifying cases as mild, moderate, or severe based on established clinical criteria, such as the Carpentier classification. 19 The integration of deep learning into echocardiographic analysis offers several advantages. AI systems can rapidly process large volumes of imaging data, reducing the workload for healthcare professionals and enabling faster diagnosis. These models also provide consistent and objective interpretations, minimizing the variability that can occur between different human observers. Importantly, AI-powered echocardiography has the potential to expand access to high-quality cardiac diagnostics, particularly in settings where expert interpretation is limited. 30 , 31 Beyond diagnostic accuracy, the integration of AI into echocardiographic screening for MR detection offers significant potential for cost-effectiveness, especially in low- and middle-income countries (LMICs). 32 AI-assisted imaging can reduce the need for specialized personnel, lower operational costs, and enable broader access through portable devices, making large-scale screening more feasible. Additionally, incorporating AI may streamline clinical workflows by automating image interpretation, allowing healthcare professionals to focus on patient management and potentially reducing wait times. 33 Despite promising results in research environments, several obstacles hinder the clinical integration of AI models. Key challenges include the need for extensive, diverse datasets to improve model generalizability, regulatory compliance, and the establishment of standardized frameworks for clinical implementation. 34 Additionally, issues such as algorithm transparency, clinician confidence, and AI interpretability must be addressed to facilitate widespread adoption. In resource-limited settings, where access to echocardiography machines and expert interpretation is scarce, AI-powered diagnostic tools could significantly impact healthcare delivery. 35 Study by Ahmed et al., identified six primary barriers to AI implementation: ethical considerations, technological constraints, regulatory and liability challenges, workforce-related issues, societal acceptance, and patient safety concerns. Addressing these challenges necessitates the standardization of reporting and performance metrics to enhance the clinical applicability of AI-driven diagnostic models. Future studies should emphasize methodological rigor, including comprehensive descriptions of study design, patient demographics, data collection methods, reference standards, and performance measures. 36 This meta-analysis represents the first comprehensive evaluation of AI algorithms’ accuracy in detecting MR compared to expert echocardiography assessments. It highlights the potential of advanced data analytics to enhance diagnostic capabilities, particularly in regions with a shortage of trained healthcare professionals needed for diagnosing complex conditions. Nevertheless, this study has certain limitations. The small number of eligible studies prevented subgroup analyses for different machine-learning techniques. Additionally, the variability in diagnostic modalities used for AI model training presents a notable challenge. Future research should focus on validating AI performance across diverse populations and echocardiographic techniques to ensure broader clinical applicability. Conclusion This meta-analysis demonstrates that AI-assisted echocardiography offers substantial potential to enhance the detection and grading of MR severity, particularly by improving diagnostic accuracy and efficiency. The adoption of advanced AI algorithms can help bridge gaps in cardiovascular care, especially in resource-limited settings where specialist expertise and imaging infrastructure are scarce. Moving forward, research should prioritize addressing these challenges, refining AI models, and developing comprehensive training and governance frameworks. With these efforts, AI-driven echocardiography could become a transformative tool in the early diagnosis and management of MR, ultimately improving patient outcomes on a global scale. Declarations Ethics approval and consent statement Not applicable. Funding statement The authors did not receive any funding for this paper. Author Contribution All authors conceived of the presented idea with LDL approval as a cardiologist. NGD and HML developed the method, independently screened, and extracted data. NGD did data statistics with support from HML and YN. LDL and YN verified the analytical methods. NGD wrote our draft manuscript. All authors discussed the results and contributed to the final manuscript. LDL is responsible for the publication process. References Figlioli G, Sticchi A, Christodoulou MN, Hadjidemetriou A, Amorim G, Alves M, et al. Global Prevalence of Mitral Regurgitation: A Systematic Review and Meta-Analysis of Population-Based Studies. Journal of Clinical Medicine 2025, Vol 14, Page 2749 [Internet]. 2025 Apr 16 [cited 2025 Apr 27];14(8):2749. Available from: https://www.mdpi.com /2077-0383/14/8/2749/htm Santangelo G, Bursi F, Faggiano A, Moscardelli S, Simeoli PS, Guazzi M, et al. The Global Burden of Valvular Heart Disease: From Clinical Epidemiology to Management. Vol. 12, Journal of Clinical Medicine. 2023. Essop MR, Nkomo VT. Rheumatic and nonrheumatic valvular heart disease: Epidemiology, management, and prevention in Africa. Circulation. 2005;112(23). Zinoviev R, Hasan RK, Gammie JS, Resar JR, Czarny MJ. Economic Burden of Inpatient Care for Mitral Regurgitation in Maryland. J Am Heart Assoc. 2024;13(2). Baumgartner H, Falk V, Bax JJ, De Bonis M, Hamm C, Holm PJ, et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J. 2017;38(36). Otto CM, Nishimura RA, Bonow RO, Carabello BA, Erwin JP, Gentile F, et al. 2020 ACC/AHA guideline for the management of patients with valvular heart disease. J Thorac Cardiovasc Surg. 2021;162(2). Malik SB, Chen N, Parker RA, Hsu JY. Transthoracic echocardiography: Pitfalls and limitations as delineated at cardiac CT and MR imaging. Radiographics. 2017;37(2). Colquhoun SM, Carapetis JR, Kado JH, Reeves BM, Remenyi B, May W, et al. Pilot study of nurse-led rheumatic heart disease echocardiography screening in Fiji - A novel approach in a resource-poor setting. Cardiol Young. 2013;23(4). Engelman D, Kado JH, Reményi B, Colquhoun SM, Carapetis JR, Donath S, et al. Focused cardiac ultrasound screening for rheumatic heart disease by briefly trained health workers: A study of diagnostic accuracy. Lancet Glob Health. 2016;4(6). Lu JC, Sable C, Ensing GJ, Webb C, Scheel J, Aliku T, et al. Simplified rheumatic heart disease screening criteria for handheld echocardiography. Journal of the American Society of Echocardiography. 2015;28(4). Brown K, Roshanitabrizi P, Rwebembera J, Okello E, Beaton A, Linguraru MG, et al. Using Artificial Intelligence for Rheumatic Heart Disease Detection by Echocardiography: Focus on Mitral Regurgitation. J Am Heart Assoc. 2024;13(2). Deeks JJ, Bossuyt PM, Leeflang MM, Takwoingi Y. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy. 2023. Whiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. Quadas-2: A revised tool for the quality assessment of diagnostic accuracy studies. Vol. 155, Annals of Internal Medicine. 2011. Long A, Haggerty CM, Finer J, Hartzel D, Jing L, Keivani A, et al. Deep Learning for Echo Analysis, Tracking, and Evaluation of Mitral Regurgitation (DELINEATE-MR). Circulation [Internet]. 2024 Sep 17 [cited 2025 Apr 29];150:911–22. Available from: /doi/pdf/10.1161/CIRCULATIONAHA.124.068996?download = true Vrudhula A, Duffy G, Vukadinovic M, Liang D, Cheng S, Ouyang D. High-Throughput Deep Learning Detection of Mitral Regurgitation. Circulation [Internet]. 2024 Sep 17 [cited 2025 Apr 29]; Available from: /doi/pdf/10.1161/CIRCULATIONAHA.124.069047?download = true Edwards LA, Feng F, Iqbal M, Fu Y, Sanyahumbi A, Hao S, et al. Machine Learning for Pediatric Echocardiographic Mitral Regurgitation Detection. Journal of the American Society of Echocardiography [Internet]. 2023 Jan 1 [cited 2025 Apr 29];36(1):96–104.e4. Available from: https://pubmed.ncbi.nlm.nih.gov/36191670/ Yang F, Chen X, Lin X, Chen X, Wang W, Liu B, et al. Automated Analysis of Doppler Echocardiographic Videos as a Screening Tool for Valvular Heart Diseases. JACC Cardiovasc Imaging. 2022;15(4). Jin CN, Salgo IS, Schneider RJ, Kam KKH, Chi WK, So CY, et al. Using Anatomic Intelligence to Localize Mitral Valve Prolapse on Three-Dimensional Echocardiography. Journal of the American Society of Echocardiography. 2016;29(10). Sadeghpour A, Jiang Z, Hummel YM, Frost M, Lam CSP, Shah SJ, et al. An Automated Machine Learning–Based Quantitative Multiparametric Approach for Mitral Regurgitation Severity Grading. JACC Cardiovasc Imaging [Internet]. 2024 Jan 1 [cited 2025 Apr 29];18(1). Available from: https://pubmed.ncbi.nlm.nih.gov/39152959/ Zhang Q, Liu Y, Mi J, Wang X, Liu X, Zhao F, et al. Automatic Assessment of Mitral Regurgitation Severity Using the Mask R-CNN Algorithm with Color Doppler Echocardiography Images. Comput Math Methods Med. 2021;2021. Moghaddasi H, Nourian S. Automatic assessment of mitral regurgitation severity based on extensive textural features on 2D echocardiography videos. Comput Biol Med. 2016;73. Douedi S, Douedi H. Mitral Regurgitation [Internet]. Statpearls. 2024 [cited 2024 Nov 29]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK553135/ Gandhi S, Mosleh W, Shen J, Chow CM. Automation, machine learning, and artificial intelligence in echocardiography: A brave new world. Vol. 35, Echocardiography. 2018. Diller GP, Kempny A, Babu-Narayan S V., Henrichs M, Brida M, Uebing A, et al. Machine learning algorithms estimating prognosis and guiding therapy in adult congenital heart disease: Data from a single tertiary centre including 10 019 patients. Eur Heart J. 2019;40(13). Liastuti LD, Nursakina Y. Diagnostic accuracy of artificial intelligence models in detecting congenital heart disease in the second-trimester fetus through prenatal cardiac screening: a systematic review and meta-analysis. Front Cardiovasc Med [Internet]. 2025 [cited 2025 Apr 29];12:1473544. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11891181/ Hoodbhoy Z, Jiwani U, Sattar S, Salam R, Hasan B, Das JK. Diagnostic Accuracy of Machine Learning Models to Identify Congenital Heart Disease: A Meta-Analysis. Vol. 4, Frontiers in Artificial Intelligence. 2021. Ghorbani A, Ouyang D, Abid A, He B, Chen JH, Harrington RA, et al. Deep learning interpretation of echocardiograms. NPJ Digit Med. 2020;3(1). Tromp J, Bauer D, Claggett BL, Frost M, Iversen MB, Prasad N, et al. A formal validation of a deep learning-based automated workflow for the interpretation of the echocardiogram. Nat Commun. 2022;13(1). Nazar W, Nazar K, Daniłowicz-Szymanowicz L. Machine Learning and Deep Learning Methods for Fast and Accurate Assessment of Transthoracic Echocardiogram Image Quality. Life [Internet]. 2024 Jun 1 [cited 2025 Apr 29];14(6):761. Available from: https://www.mdpi.com/ 2075-1729/14/6/761/htm Hirata Y, Kusunose K. AI in Echocardiography: State-of-the-art Automated Measurement Techniques and Clinical Applications. JMA J [Internet]. 2024 [cited 2025 Apr 29];8(1):141. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11799715/ Jiang L, Zuo HJ, Chen C. Artificial intelligence in echocardiography: Applications and future directions. Fundamental Research [Internet]. 2025 Apr 25 [cited 2025 Apr 29]; Available from: https://linkinghub.elsevier.com/retrieve/pii/S2667325825001025 Kusunose K. Transforming Echocardiography: The Role of Artificial Intelligence in Enhancing Diagnostic Accuracy and Accessibility. Internal Medicine [Internet]. 2024 Feb 1 [cited 2025 Apr 29];64(3):331. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11867741/ Alabdaljabar MS, Hasan B, Noseworthy PA, Maalouf JF, Ammash NM, Hashmi SK. Machine Learning in Cardiology: A Potential Real-World Solution in Low- and Middle-Income Countries. J Multidiscip Healthc [Internet]. 2023 [cited 2025 Apr 29];16:285–95. Available from: https://www.tandfonline.com/doi/pdf/ 10.2147/JMDH.S383810 He J, Baxter SL, Xu J, Xu J, Zhou X, Zhang K. The practical implementation of artificial intelligence technologies in medicine. Vol. 25, Nature Medicine. 2019. Dey D, Slomka PJ, Leeson P, Comaniciu D, Shrestha S, Sengupta PP, et al. Artificial Intelligence in Cardiovascular Imaging: JACC State-of-the-Art Review. Vol. 73, Journal of the American College of Cardiology. 2019. Ahmed MI, Spooner B, Isherwood J, Lane M, Orrock E, Dennison A. A Systematic Review of the Barriers to the Implementation of Artificial Intelligence in Healthcare. Cureus. 2023; Additional Declarations No competing interests reported. Supplementary Files SupplementaryTable.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6626740","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":458322240,"identity":"0e3541db-25fc-410d-b628-0f26fce27319","order_by":0,"name":"Lies Dina Liastuti","email":"data:image/png;base64,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","orcid":"","institution":"Pembuluh Darah Harapan Kita","correspondingAuthor":true,"prefix":"","firstName":"Lies","middleName":"Dina","lastName":"Liastuti","suffix":""},{"id":458322241,"identity":"2e041e01-f66e-49ea-b169-1ecaec594028","order_by":1,"name":"Nathaniel Gilbert Dyson","email":"","orcid":"","institution":"University of Indonesia","correspondingAuthor":false,"prefix":"","firstName":"Nathaniel","middleName":"Gilbert","lastName":"Dyson","suffix":""},{"id":458322242,"identity":"d2bdaf3f-9730-480e-bc0e-2548933f52b9","order_by":2,"name":"Haifa Mayang Lestari","email":"","orcid":"","institution":"University of Indonesia","correspondingAuthor":false,"prefix":"","firstName":"Haifa","middleName":"Mayang","lastName":"Lestari","suffix":""},{"id":458322243,"identity":"fb183ec3-3632-4fdd-aa10-dd613e2ba88b","order_by":3,"name":"Yosilia Nursakina","email":"","orcid":"","institution":"University of Indonesia","correspondingAuthor":false,"prefix":"","firstName":"Yosilia","middleName":"","lastName":"Nursakina","suffix":""}],"badges":[],"createdAt":"2025-05-09 08:38:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6626740/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6626740/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":83109200,"identity":"11ad9233-16c2-41a8-b3f2-992d241ad8bb","added_by":"auto","created_at":"2025-05-20 06:54:27","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":241389,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA flowchart of included studies\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/6a80b0d038d9266d0db02aa6.jpg"},{"id":83109199,"identity":"f232efd0-1cc1-4b7e-984e-0c02ace9d9c7","added_by":"auto","created_at":"2025-05-20 06:54:27","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":157487,"visible":true,"origin":"","legend":"\u003cp\u003eRisk of bias assessment using QUADAS-2 tools\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/6e998cbc6899ebd09ddddf8b.jpg"},{"id":83108142,"identity":"73d04793-55a8-48c6-b5ff-f21500626746","added_by":"auto","created_at":"2025-05-20 06:46:27","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":51130,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot for MR detection sensitivity\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/71acdcee9a371a8321e56538.jpg"},{"id":83109206,"identity":"2e6fc473-0209-497e-9f67-f98526e599da","added_by":"auto","created_at":"2025-05-20 06:54:27","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":52117,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot for accuracy of MR detection specificity\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/be49fbd0904f464d97510e5d.jpg"},{"id":83108141,"identity":"fa10383a-309c-4dc5-ab4d-e731e5577fdf","added_by":"auto","created_at":"2025-05-20 06:46:27","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":16018,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot for SROC curve for MR detection\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/0368df7681aa6acf7fac83b8.jpg"},{"id":109405610,"identity":"2d2476ee-5607-4b95-b543-86edf6a60f6c","added_by":"auto","created_at":"2026-05-17 13:19:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":786572,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/5530c79c-a25b-4495-b08f-1a0f31d69462.pdf"},{"id":83108139,"identity":"dd83a631-7e8e-4f49-907d-f270a7b7fe90","added_by":"auto","created_at":"2025-05-20 06:46:27","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":17368,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable.docx","url":"https://assets-eu.researchsquare.com/files/rs-6626740/v1/e052a846e4682848fda5f2d1.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrating Artificial Intelligence for Mitral Regurgitation Assessment: A systematic review and meta-analysis","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMitral regurgitation (MR) is one of the most prevalent valvular heart diseases worldwide and represents a significant contributor to cardiovascular morbidity and mortality. Recent data suggest that the global prevalence of moderate-to-severe MR is approximately 0.67% (95% CI, 0.33\u0026ndash;1.11), based on population-based studies including more than 6\u0026nbsp;million individuals.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e The clinical impact of MR is substantial, as it often leads to symptoms such as exertional dyspnea and fatigue, particularly in advanced stages, thereby markedly diminishing patients\u0026rsquo; quality of life.\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e In regions where RHD remains endemic, MR is a common sequela, often affecting younger individuals and contributing substantially to early morbidity and mortality.\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eMR has serious financial repercussions for healthcare systems in addition to its clinical effects. According to recent data from Maryland, USA, individuals undergoing medical therapy for MR had to pay an average of \u003cspan\u003e$\u003c/span\u003e23,575 a year. These patients spent 3.1 more days in the hospital annually and had higher annual healthcare expenses (\u003cspan\u003e$\u003c/span\u003e10,559) than those without MR. The average expenses for patients having transcatheter or surgical mitral valve replacement or repair were \u003cspan\u003e$\u003c/span\u003e63,108 and \u003cspan\u003e$\u003c/span\u003e47,943 for the treatment year, respectively. In contrast to continuing medical care, yearly costs for both groups were much lower in the second and third years after these treatments.\u003csup\u003e \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e \u003c/sup\u003e \u003c/p\u003e \u003cp\u003eSince the underlying cause has a significant influence on both treatment choices and long-term results, determining the MR etiology accurately is essential for directing suitable management techniques. Infectious endocarditis, rheumatic heart disease, coronary artery disease, MVP, flail leaflets, drug-induced valvulopathy, and connective tissue diseases are among the intrinsic defects of the mitral valve apparatus that usually cause primary MR. While chronic MR frequently develops gradually over time, acute severe MR can result from abrupt structural breakdown, such as chordae tendineae rupture or papillary muscle dysfunction.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e According to current clinical guidelines, patients with suspected MR should have transthoracic echocardiography for diagnosis and then routine echocardiographic monitoring at the right intervals to avoid irreversible damage to the pulmonary circulation and ventricles, which can happen without cause.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e MR can be diagnosed noninvasively using auscultation or patient interview, however neither technique is objective, and examiner accuracy varies greatly. Even though echocardiography is the gold standard, doing it takes a lot of time and skill.\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e Therefore, more objective techniques to assess echocardiographic imaging of MR patients are desperately needed.\u003c/p\u003e \u003cp\u003eDespite the recognized importance of early diagnosis, the implementation of widespread echocardiographic screening programs in low- and middle-income countries (LMICs) remains a significant challenge. These regions, which bear a disproportionate burden of rheumatic heart disease, often lack adequate imaging infrastructure, specialized personnel, and standardized protocols.\u003csup\u003e\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e Various strategies have been proposed to address these barriers, including the deployment of portable ultrasound devices, simplification of imaging workflows, and the training of non-physician healthcare providers.\u003c/p\u003e \u003cp\u003eConcurrently, rapid advancements in artificial intelligence (AI) have created new opportunities to enhance cardiac imaging. AI-powered systems are being developed to automate both image acquisition and interpretation, potentially reducing operator dependency and increasing diagnostic consistency across diverse clinical settings.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e These innovations are particularly relevant in resource-limited environments, where access to trained echocardiographers is restricted.\u003c/p\u003e \u003cp\u003eGiven the increasing global burden of MR and the limitations of conventional diagnostic approaches, AI-assisted tools may offer a scalable and efficient alternative for early detection and risk stratification. This review aims to answer whether AI-based echocardiographic assessment can match or surpass traditional methods in accuracy and reliability, thereby supporting wider adoption in diverse healthcare settings. By synthesizing current evidence, this review seeks to inform clinical guidelines and support the equitable integration of AI technologies into cardiovascular care.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eInformation sources and search strategy\u003c/h2\u003e \u003cp\u003eThis systematic review and meta-analysis was conducted following a pre-registered protocol with the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42024552951). To find pertinent papers published up to April 27, 2025, a thorough literature search was conducted across five major electronic databases: PubMed, Cochrane Library, ProQuest, Scopus, and ScienceDirect. In addition to free-text terms like \"Mitral Valve Insufficiency,\" \"Artificial Intelligence,\" and \"Machine Learning,\" the search strategy combined Medical Subject Headings (MeSH) with echocardiography-specific terms like \"Echocardiography, Three-Dimensional\" and \"Echocardiography, Transesophageal.\" This strategy was created to guarantee that all relevant material was included and to optimize the search's sensitivity and specificity.\u003c/p\u003e \u003cp\u003eDuplicate entries were found and eliminated once all recovered records were loaded into the EndNote X9 reference management program. To determine study eligibility, two reviewers (NG and HM) independently evaluated abstracts and titles. The final inclusion of articles that were judged possibly relevant was decided by full-text examination. A third reviewer was consulted or spoken with in order to settle any disputes that arose throughout the screening or selection process. At the full-text stage, the reasons for exclusion were also well recorded. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy's methodological requirements were followed in the conduct of this review.\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTable of keywords for literature search\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDatabase\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSearch strategy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHits\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePubmed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(\"Mitral Valve Insufficiency\"[Mesh] OR \"Mitral Valve Insufficiency/diagnosis\"[Mesh] OR \"Mitral Valve Insufficiency/diagnostic imaging\"[Mesh]) AND ((((\"Artificial Intelligence\"[Mesh]) OR \"Machine Learning\"[Mesh]) OR \"Supervised Machine Learning\"[Mesh]) OR \"Support Vector Machine\"[Mesh]) OR \"Unsupervised Machine Learning\"[Mesh] AND (\"Echocardiography\"[Mesh] OR \"Echocardiography, Stress\"[Mesh] OR \"Echocardiography, Four-Dimensional\"[Mesh] OR \"Echocardiography, Three-Dimensional\"[Mesh] OR \"Echocardiography, Doppler, Pulsed\"[Mesh] OR \"Echocardiography, Doppler, Color\"[Mesh] OR \"Echocardiography, Transesophageal\"[Mesh] OR \"Echocardiography, Doppler\"[Mesh])\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e57 results\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCochrane\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(mitral valve regurgitation OR mitral valve insufficiency OR mitral regurgitation in Title Abstract Keyword) AND (artificial intelligence OR machine learning OR supervised machine learning OR unsupervised machine learning OR deep learning in Title Abstract Keywords) AND (echocardiography OR echocardiogram in Title Abstract Keywords)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 results\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eProquest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(Mitral Valve Insufficiency OR Mitral Regurgitation OR Mitral Valve Regurgitation) AND (Artificial Intelligence OR Machine Learning OR Supervised Machine Learning OR Support Vector Machine OR Unsupervised Machine Learning) AND (echocardiography) AND (accuracy OR area under the curve OR AUC OR sensitivity OR specificity) AND (heart function OR ventricular function)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1,284 results\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScopus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTITLE-ABS-KEY ( ( mitral AND valve AND insufficiency OR mitral AND regurgitation OR mitral AND valve AND regurgitation ) AND ( artificial AND intelligence OR machine AND learning OR supervised AND machine AND learning OR support AND vector AND machine OR unsupervised AND machine AND learning ) AND ( echocardiography ) )\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10 results\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScience Direct\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(\"Mitral Valve Insufficiency\" OR \"Mitral Regurgitation\" OR \"Mitral Valve Regurgitation\") AND (\"Artificial Intelligence\" OR \"Machine Learning\" OR \"Supervised Machine Learning\" OR \"Support Vector Machine\" OR Unsupervised Machine Learning) AND \"echocardiography\"\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e265 results\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy eligibility criteria\u003c/h3\u003e\n\u003cp\u003eThis review included observational studies\u0026mdash;specifically cross-sectional and cohort designs\u0026mdash;that investigated the diagnostic accuracy of AI or machine learning algorithms in identifying MR. Eligible studies were required to utilize a recognized reference standard, such as interpretation from echocardiographers or cardiologists for MR diagnosis. Inclusion was limited to articles published in English. There were no restrictions regarding patient age at diagnosis or the specific AI methodology applied. Studies that did not clearly define the reference standard were excluded. For studies that assessed multiple types of valvular heart disease, only data pertaining to MR were extracted and analyzed.\u003c/p\u003e\n\u003ch3\u003eData extraction\u003c/h3\u003e\n\u003cp\u003eTwo independent reviewers (NG and HM) performed data extraction using a standardized, pretested form to ensure methodological rigor and consistency. Extracted information included bibliographic details (author names, publication year, and study title), geographical location, study design, clinical setting, and sample size\u0026mdash;distinguishing between training and testing datasets where applicable. Patient selection criteria were also recorded. In addition, technical characteristics of the AI/ML models were documented, including model architecture, input data type, reference standard used for model validation (e.g., expert echocardiographers or cardiologists), diagnostic performance indicators (e.g., sensitivity, specificity, area under the curve [AUC]), and details of internal or external validation procedures. Any disagreements regarding study inclusion, data abstraction, or interpretation were resolved through iterative discussion. If consensus could not be achieved, a third reviewer (LD) was consulted to provide adjudication. This triangulated review process ensured methodological transparency and reduced the risk of bias.\u003c/p\u003e\n\u003ch3\u003eQuality assessment\u003c/h3\u003e\n\u003cp\u003eThe Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) instrument was used by two reviewers (NG and HM) to independently evaluate the risk of bias.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e Four areas are assessed by this framework: flow and timing, reference standard, index test, and patient selection. Concerns about applicability were also evaluated for the first three areas. Any item inside each domain that was deemed to be high risk resulted in the domain as a whole being classified as high risk.Consensus talks were used to settle disagreements on risk of bias assessments. A third reviewer (LD) acted as an arbitrator to complete evaluations when needed. A thorough and objective evaluation of the study's quality was guaranteed by this methodical and planned methodology.\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eQuantitative synthesis of diagnostic performance was performed using Meta-DiSc version 1.4. For each included study, pooled estimates of sensitivity and specificity were calculated, along with their corresponding 95% confidence intervals (CIs). Summary receiver operating characteristic (SROC) curves were generated to provide an overall measure of diagnostic accuracy, and results were visually presented using forest plots and SROC curves. To investigate potential sources of heterogeneity, subgroup analyses were conducted, with particular attention to the types of artificial intelligence (AI) or machine learning (ML) algorithms utilized. Where appropriate, sensitivity analyses were carried out to evaluate the robustness of the pooled estimates and to determine the impact of individual studies on the overall results.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStudy characteristics\u003c/h2\u003e \u003cp\u003eThe initial search yielded 1,618 articles, of which 1,555 were deemed eligible for screening after removing 63 duplicates. Following title and abstract screening, 1,278 studies were excluded due to irrelevant title. Of the remaining 277 studies, 246 were further excluded due to irrelevant abstract screening, leaving 31 studies for full-text review. After detailed evaluation, 20 studies were excluded: 7 were not original research, and 15 lacked accessible full-text versions. Finally, 9 studies met the inclusion criteria and were included in the review (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e for the study selection flowchart).\u003c/p\u003e \u003cp\u003eThe characteristics of the nine studies included in this review are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. All studies employed were observational studies, with seven adopting a cross-sectional approach and two utilizing retrospective cohort methodologies. Cross-sectional studies assessed the diagnostic performance of AI algorithms at a single time point, while retrospective cohort studies enabled evaluation of algorithm performance over previously collected datasets. Specifically, five studies were performed in high-income countries (four in the USA and one in Hong Kong), while three were conducted in middle-income countries (two in China and one in Iran). The included studies analyzed echocardiographic images from diverse datasets. Various AI methodologies were employed, including convolutional neural networks (CNN), deep learning (DL), support vector machines (SVM), machine learning (ML), and artificial intelligence-based ultrasound systems (AIUS). We further categorized the included studies based on their primary objective, distinguishing between those that evaluated the diagnostic accuracy of AI algorithms for MR detection and those that focused on grading the severity of mitral regurgitation.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eBias assessment\u003c/h3\u003e\n\u003cp\u003eThe risk of bias assessment for the included studies is summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Overall, the majority of studies demonstrated a low risk of bias across all evaluated domains. Most studies were rated as low risk for patient selection, index test, reference standard, and flow and timing. However, some concerns were noted in specific domains for a few studies. Zhang et al. (2021) and Moghaddasi et al. (2016) showed some concerns regarding the reference standard, while Brown et al. (2024), Long et al. (2024), Zhang et al. (2021), and Moghaddasi et al. (2016) had some concerns in the flow and timing domain. Despite these minor concerns, no issues were identified regarding the applicability of the index test or patient selection. These findings suggest that the included studies are methodologically sound and that their results are relevant and appropriately aligned with the objectives of this review.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eAI algorithms and reference standards\u003c/h2\u003e \u003cp\u003eA variety of artificial intelligence (AI) algorithms were utilized across the included studies, reflecting the evolving landscape of machine learning applications in cardiovascular imaging. Convolutional neural networks (CNNs) and deep learning (DL) models were the most frequently employed, particularly for tasks involving MR detection and image-based classification. For example, Brown et al. (2024) and Edwards et al. (2023) implemented CNNs for automated MR detection, while Long et al. (2024) and Vrudhula et al. (2024) used broader DL approaches with large-scale echocardiographic datasets. Support vector machines (SVMs) were also applied, notably in studies by Yang et al. (2022) and Moghaddasi (2016), primarily for MR severity grading and classification tasks. Additionally, other machine learning (ML) techniques and specialized algorithms such as artificial intelligence-based ultrasound systems (AIUS) were reported, as seen in Jin et al. (2016) and Sadeghpour et al. (2025).\u003c/p\u003e \u003cp\u003eThe reference standards for evaluating AI performance varied among studies, reflecting differences in clinical practice and available expertise. Most studies used diagnoses established by experienced cardiologists or echocardiographers as the reference standard, either individually or through expert consensus panels. For instance, Brown et al. (2024) and Jin et al. (2016) validated their AI models against the combined assessments of echocardiographers and cardiologists, while others, such as Long et al. (2024) and Vrudhula et al. (2024), relied on cardiologist adjudication alone. In some cases, expert echocardiographers provided detailed grading of MR severity, as in Zhang et al. (2021) and Moghaddasi (2016).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of study characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStudy Design\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIncome region\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTraining dataset\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eTesting dataset\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eAI Algorithm\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"7\" nameend=\"c7\" namest=\"c1\"\u003e \u003cp\u003eMR Detection\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBrown et al\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e416\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLong et al\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, 2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e43,811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e17,878\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eDL\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVrudhula et al\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, 2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e80,833\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e46,890\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eDL\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEdwards et al\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e, 2023\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e66,330\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e11,730\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYang et al\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRetrospective cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eChina\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMiddle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e777\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e151\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eJin et al\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRetrospective cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHong Kong\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eAIUS\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"7\" nameend=\"c7\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMR Severity Grading\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSadeghpour et al\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUSA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e438\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e438\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eML\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZhang et al\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eChina\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMiddle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1,132\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e295\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMoghaddasi et al\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e, 2016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCross-sectional\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIran\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMiddle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e102\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e102\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of study results\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eStudy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eIndex test\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eReference standard\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c7\" namest=\"c4\"\u003e \u003cp\u003eStudy outcomes\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eF1 score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"8\" nameend=\"c8\" namest=\"c1\"\u003e \u003cp\u003eMR Detection\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBrown et al\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEchocardiographers and cardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.93\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLong et al\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.98\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVrudhula et al\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.975\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.999\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.916\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEdwards et al\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2023\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.91\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYang et al\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eJin et al\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e, 2016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAIUS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEchocardiographers and cardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"8\" nameend=\"c8\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMR Severity Grading\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSadeghpour et al\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e, 2025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eML\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCardiologists\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eMild: 0.80\u003c/p\u003e \u003cp\u003eModerate and severe: 0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZhang et al\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e,\u003c/p\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEchocardiographers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eI: 0.90\u003c/p\u003e \u003cp\u003eII: 0.87\u003c/p\u003e \u003cp\u003eIII: 0.81\u003c/p\u003e \u003cp\u003eIV: 0.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eI: 0.94\u003c/p\u003e \u003cp\u003eII: 0.88\u003c/p\u003e \u003cp\u003eIII: 0.85\u003c/p\u003e \u003cp\u003eIV: 0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMoghaddasi et al\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e, 2016\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEchocardiographers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMild: 0.99\u003c/p\u003e \u003cp\u003eModerate: 0.98\u003c/p\u003e \u003cp\u003eSevere: 0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMild: 0.99\u003c/p\u003e \u003cp\u003eModerate: 0.99\u003c/p\u003e \u003cp\u003eSevere: 0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eAI accuracy in MR detection\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOur meta-analysis demonstrated that AI-assisted echocardiographic analysis for the detection of mitral regurgitation (MR) yielded a pooled sensitivity of 0.86 (95% confidence interval [CI]: 0.86 to 0.86). This finding indicates that AI algorithms are highly effective in identifying MR across the included studies. However, substantial heterogeneity was observed, as reflected by an I\u0026sup2; value of 99.0%, far exceeding the conventional threshold of 75% for high heterogeneity. The chi-square test further confirmed this variability (χ\u0026sup2; = 505.33, df\u0026thinsp;=\u0026thinsp;5, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001). Despite the robust pooled sensitivity, the marked heterogeneity suggests that factors such as algorithm architecture, population characteristics, and imaging equipment may influence diagnostic performance. These results highlight the need for standardized evaluation protocols and further research to optimize and generalize the application of AI in echocardiographic MR detection.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOur meta-analysis showed that AI-assisted echocardiography for the detection of mitral regurgitation (MR) achieved a pooled specificity of 0.83 (95% confidence interval [CI]: 0.82 to 0.83). This suggests that AI algorithms demonstrate good accuracy in correctly identifying individuals without MR. However, substantial heterogeneity was observed among the included studies, as indicated by an I\u0026sup2; value of 99.4%, which is well above the conventional threshold for high heterogeneity. The chi-square test further supported this finding (χ\u0026sup2; = 770.74, df\u0026thinsp;=\u0026thinsp;5, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), highlighting significant variability in specificity estimates across studies.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOur meta-analysis demonstrates that AI-assisted echocardiography for the detection of mitral regurgitation (MR) achieves excellent diagnostic accuracy, as reflected by an area under the curve (AUC) of 0.9745 on the symmetric summary receiver operating characteristic (SROC) curve. The AUC is a key metric for evaluating the overall performance of diagnostic tests, with values approaching 1.0 indicating outstanding discriminative ability. The observed AUC underscores the high reliability of AI algorithms in distinguishing between patients with and without MR.\u003c/p\u003e \u003cp\u003eThe precision of this estimate is supported by a standard error (SE) of 0.0181 for the AUC. Additionally, the Q* index\u0026mdash;a summary measure of diagnostic effectiveness\u0026mdash;was 0.9274 (SE\u0026thinsp;=\u0026thinsp;0.0311), further confirming the robustness of the AI-guided approach. Collectively, these findings highlight the strong potential of AI-driven tools to enhance MR detection and support more efficient and accurate diagnostic workflows in clinical practice.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eAI accuracy in MR severity grading\u003c/h2\u003e \u003cp\u003eThe included studies demonstrate that artificial intelligence (AI) algorithms can achieve high accuracy in grading the severity of mitral regurgitation (MR) using echocardiographic data. However, due to lack of data, we could not perform meta-analysis on the MR severity grading parameter. Three studies\u0026mdash;Sadeghpour et al\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e, Zhang et al\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, and Moghaddasi et al\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e\u0026mdash;specifically evaluated the diagnostic performance of various AI models for MR severity classification. The findings from each study are displayed in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Sadeghpour et al\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e employed a machine learning (ML) approach validated by cardiologists. Their model achieved a sensitivity of 0.96 and a specificity of 0.98 for MR severity grading. The accuracy was 0.80 for mild MR and 0.97 for moderate and severe MR categories, indicating strong performance across different severity levels.\u003c/p\u003e \u003cp\u003eZhang et al\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e developed a Mask R-CNN deep learning algorithm for automatic MR severity assessment using color Doppler echocardiography images, following the 2017 American Society of Echocardiography (ASE) guidelines. The study included a large, multi-center dataset and validated the model both internally and externally. The algorithm achieved classification accuracies of 0.90 for grade I (mild), 0.87 for grade II (moderate), 0.81 for grade III (moderate), and 0.91 for grade IV (severe). The corresponding specificities were 0.94, 0.88, 0.85, and 0.89, respectively. The Macro F1 and Micro F1 scores for grading were both 0.89, reflecting balanced performance across all MR grades. The study also highlighted that the AI model substantially reduced the time required for MR severity assessment compared to manual guideline-based evaluation, and maintained robust performance across different hospitals and ultrasound equipment.\u003c/p\u003e \u003cp\u003eMoghaddasi et al\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e utilized a support vector machine (SVM) model to classify MR severity into mild, moderate, and severe categories, based on echocardiographic images. The model achieved high sensitivity and specificity for all classes: 0.99 for mild, 0.98 for moderate, and 0.99 for severe MR, with corresponding specificities of 0.99, 0.99, and 0.99, and classification accuracies exceeding 99% for each category.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eTo the best of our knowledge, this is the first meta-analysis to systematically evaluate the diagnostic accuracy of AI models for both detection and classification of MR. Mitral regurgitation is a prevalent valvular disorder characterized by the improper closure of the mitral valve, leading to the backward flow of blood into the left atrium.\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e In order to prevent consequences like atrial fibrillation and heart failure, early and accurate diagnosis is essential. However, prompt intervention is hampered by obstacles including limited access to echocardiographic facilities and expert cardiologists, especially in under-resourced areas. By allowing non-physician healthcare practitioners to do tests, AI-driven echocardiography holds promise for closing these disparities. AI combined with portable echocardiography equipment provides a scalable and economical diagnostic method, especially in settings with limited resources.\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e Our findings underscore the significant potential of AI to improve MR diagnosis, especially in environments with limited access to specialized healthcare professionals. Echocardiographic imaging emerged as the predominant data source for AI algorithms, with convolutional neural networks and support vector machines being the most commonly utilized analytical methods. Pooled results demonstrated that AI-based echocardiography achieved high sensitivity (86%) and high specificity (83%) for MR detection.\u003c/p\u003e \u003cp\u003eThe results of this meta-analysis are consistent with findings from several other recent meta-analyses evaluating the application of artificial intelligence in echocardiographic analysis. For instance, a systematic review and meta-analysis by Liastuti et al assessed the diagnostic accuracy of AI models in detecting congenital heart disease during second-trimester fetal cardiac screening, demonstrating that AI can achieve high sensitivity and specificity in this context.\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e Similarly, another meta-analysis examined the performance of machine learning models in identifying congenital heart disease, further supporting the reliability of AI-based approaches in cardiac imaging.\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eConvolutional neural networks (CNNs), a type of deep learning model, are particularly well-suited for interpreting complicated visual data, such as echocardiograms. These algorithms can automatically recognize complex patterns in the pictures, including aberrant mitral valve motion or regurgitant blood flow, that might point to the existence of MR. Usually, the procedure starts with the gathering of video clips or echocardiograms, which are subsequently preprocessed to improve picture quality and standardize the data. Normalization of pixel values, noise reduction, and segmentation of pertinent cardiac structures, such as the left atrium and mitral valve, are examples of preprocessing procedures. This guarantees that reliable and superior input data for analysis is provided to the deep learning model.\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eUnlike traditional machine learning approaches that rely on manual feature selection, deep learning models automatically extract relevant features from the raw image data. Through multiple layers of convolutional filters, CNNs can identify subtle patterns and features that are characteristic of MR. The models are trained on large datasets of labeled echocardiography images, where each image is annotated by expert cardiologists to indicate the presence or absence of MR. During training, the model learns to associate specific visual features with MR, gradually improving its diagnostic accuracy.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e Once trained, the deep learning model can analyze new echocardiography images and predict whether MR is present. Some advanced models are also capable of grading the severity of MR, classifying cases as mild, moderate, or severe based on established clinical criteria, such as the Carpentier classification.\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThe integration of deep learning into echocardiographic analysis offers several advantages. AI systems can rapidly process large volumes of imaging data, reducing the workload for healthcare professionals and enabling faster diagnosis. These models also provide consistent and objective interpretations, minimizing the variability that can occur between different human observers. Importantly, AI-powered echocardiography has the potential to expand access to high-quality cardiac diagnostics, particularly in settings where expert interpretation is limited.\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eBeyond diagnostic accuracy, the integration of AI into echocardiographic screening for MR detection offers significant potential for cost-effectiveness, especially in low- and middle-income countries (LMICs).\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e AI-assisted imaging can reduce the need for specialized personnel, lower operational costs, and enable broader access through portable devices, making large-scale screening more feasible. Additionally, incorporating AI may streamline clinical workflows by automating image interpretation, allowing healthcare professionals to focus on patient management and potentially reducing wait times.\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eDespite promising results in research environments, several obstacles hinder the clinical integration of AI models. Key challenges include the need for extensive, diverse datasets to improve model generalizability, regulatory compliance, and the establishment of standardized frameworks for clinical implementation.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e Additionally, issues such as algorithm transparency, clinician confidence, and AI interpretability must be addressed to facilitate widespread adoption. In resource-limited settings, where access to echocardiography machines and expert interpretation is scarce, AI-powered diagnostic tools could significantly impact healthcare delivery.\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e Study by Ahmed et al., identified six primary barriers to AI implementation: ethical considerations, technological constraints, regulatory and liability challenges, workforce-related issues, societal acceptance, and patient safety concerns. Addressing these challenges necessitates the standardization of reporting and performance metrics to enhance the clinical applicability of AI-driven diagnostic models. Future studies should emphasize methodological rigor, including comprehensive descriptions of study design, patient demographics, data collection methods, reference standards, and performance measures.\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThis meta-analysis represents the first comprehensive evaluation of AI algorithms\u0026rsquo; accuracy in detecting MR compared to expert echocardiography assessments. It highlights the potential of advanced data analytics to enhance diagnostic capabilities, particularly in regions with a shortage of trained healthcare professionals needed for diagnosing complex conditions. Nevertheless, this study has certain limitations. The small number of eligible studies prevented subgroup analyses for different machine-learning techniques. Additionally, the variability in diagnostic modalities used for AI model training presents a notable challenge. Future research should focus on validating AI performance across diverse populations and echocardiographic techniques to ensure broader clinical applicability.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis meta-analysis demonstrates that AI-assisted echocardiography offers substantial potential to enhance the detection and grading of MR severity, particularly by improving diagnostic accuracy and efficiency. The adoption of advanced AI algorithms can help bridge gaps in cardiovascular care, especially in resource-limited settings where specialist expertise and imaging infrastructure are scarce. Moving forward, research should prioritize addressing these challenges, refining AI models, and developing comprehensive training and governance frameworks. With these efforts, AI-driven echocardiography could become a transformative tool in the early diagnosis and management of MR, ultimately improving patient outcomes on a global scale.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding statement\u003c/h2\u003e \u003cp\u003eThe authors did not receive any funding for this paper.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAll authors conceived of the presented idea with LDL approval as a cardiologist. NGD and HML developed the method, independently screened, and extracted data. NGD did data statistics with support from HML and YN. LDL and YN verified the analytical methods. NGD wrote our draft manuscript. All authors discussed the results and contributed to the final manuscript. LDL is responsible for the publication process.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eFiglioli G, Sticchi A, Christodoulou MN, Hadjidemetriou A, Amorim G, Alves M, et al. Global Prevalence of Mitral Regurgitation: A Systematic Review and Meta-Analysis of Population-Based Studies. Journal of Clinical Medicine 2025, Vol 14, Page 2749 [Internet]. 2025 Apr 16 [cited 2025 Apr 27];14(8):2749. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.mdpi.com\u003c/span\u003e\u003cspan address=\"https://www.mdpi.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e/2077-0383/14/8/2749/htm\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSantangelo G, Bursi F, Faggiano A, Moscardelli S, Simeoli PS, Guazzi M, et al. The Global Burden of Valvular Heart Disease: From Clinical Epidemiology to Management. Vol. 12, Journal of Clinical Medicine. 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEssop MR, Nkomo VT. Rheumatic and nonrheumatic valvular heart disease: Epidemiology, management, and prevention in Africa. Circulation. 2005;112(23).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZinoviev R, Hasan RK, Gammie JS, Resar JR, Czarny MJ. Economic Burden of Inpatient Care for Mitral Regurgitation in Maryland. J Am Heart Assoc. 2024;13(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaumgartner H, Falk V, Bax JJ, De Bonis M, Hamm C, Holm PJ, et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J. 2017;38(36).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOtto CM, Nishimura RA, Bonow RO, Carabello BA, Erwin JP, Gentile F, et al. 2020 ACC/AHA guideline for the management of patients with valvular heart disease. J Thorac Cardiovasc Surg. 2021;162(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalik SB, Chen N, Parker RA, Hsu JY. Transthoracic echocardiography: Pitfalls and limitations as delineated at cardiac CT and MR imaging. Radiographics. 2017;37(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eColquhoun SM, Carapetis JR, Kado JH, Reeves BM, Remenyi B, May W, et al. Pilot study of nurse-led rheumatic heart disease echocardiography screening in Fiji - A novel approach in a resource-poor setting. Cardiol Young. 2013;23(4).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEngelman D, Kado JH, Rem\u0026eacute;nyi B, Colquhoun SM, Carapetis JR, Donath S, et al. Focused cardiac ultrasound screening for rheumatic heart disease by briefly trained health workers: A study of diagnostic accuracy. Lancet Glob Health. 2016;4(6).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu JC, Sable C, Ensing GJ, Webb C, Scheel J, Aliku T, et al. Simplified rheumatic heart disease screening criteria for handheld echocardiography. Journal of the American Society of Echocardiography. 2015;28(4).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrown K, Roshanitabrizi P, Rwebembera J, Okello E, Beaton A, Linguraru MG, et al. Using Artificial Intelligence for Rheumatic Heart Disease Detection by Echocardiography: Focus on Mitral Regurgitation. J Am Heart Assoc. 2024;13(2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeeks JJ, Bossuyt PM, Leeflang MM, Takwoingi Y. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy. 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWhiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. Quadas-2: A revised tool for the quality assessment of diagnostic accuracy studies. Vol. 155, Annals of Internal Medicine. 2011.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong A, Haggerty CM, Finer J, Hartzel D, Jing L, Keivani A, et al. Deep Learning for Echo Analysis, Tracking, and Evaluation of Mitral Regurgitation (DELINEATE-MR). Circulation [Internet]. 2024 Sep 17 [cited 2025 Apr 29];150:911\u0026ndash;22. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e/doi/pdf/10.1161/CIRCULATIONAHA.124.068996?download\u0026thinsp;=\u0026thinsp;true\u003c/span\u003e\u003cspan address=\"/doi/pdf/10.1161/CIRCULATIONAHA.124.068996?download\u0026thinsp;=\u0026thinsp;true\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVrudhula A, Duffy G, Vukadinovic M, Liang D, Cheng S, Ouyang D. High-Throughput Deep Learning Detection of Mitral Regurgitation. Circulation [Internet]. 2024 Sep 17 [cited 2025 Apr 29]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e/doi/pdf/10.1161/CIRCULATIONAHA.124.069047?download\u0026thinsp;=\u0026thinsp;true\u003c/span\u003e\u003cspan address=\"/doi/pdf/10.1161/CIRCULATIONAHA.124.069047?download\u0026thinsp;=\u0026thinsp;true\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEdwards LA, Feng F, Iqbal M, Fu Y, Sanyahumbi A, Hao S, et al. Machine Learning for Pediatric Echocardiographic Mitral Regurgitation Detection. Journal of the American Society of Echocardiography [Internet]. 2023 Jan 1 [cited 2025 Apr 29];36(1):96\u0026ndash;104.e4. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/36191670/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/36191670/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang F, Chen X, Lin X, Chen X, Wang W, Liu B, et al. Automated Analysis of Doppler Echocardiographic Videos as a Screening Tool for Valvular Heart Diseases. JACC Cardiovasc Imaging. 2022;15(4).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJin CN, Salgo IS, Schneider RJ, Kam KKH, Chi WK, So CY, et al. Using Anatomic Intelligence to Localize Mitral Valve Prolapse on Three-Dimensional Echocardiography. Journal of the American Society of Echocardiography. 2016;29(10).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSadeghpour A, Jiang Z, Hummel YM, Frost M, Lam CSP, Shah SJ, et al. An Automated Machine Learning\u0026ndash;Based Quantitative Multiparametric Approach for Mitral Regurgitation Severity Grading. JACC Cardiovasc Imaging [Internet]. 2024 Jan 1 [cited 2025 Apr 29];18(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/39152959/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/39152959/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Q, Liu Y, Mi J, Wang X, Liu X, Zhao F, et al. Automatic Assessment of Mitral Regurgitation Severity Using the Mask R-CNN Algorithm with Color Doppler Echocardiography Images. Comput Math Methods Med. 2021;2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoghaddasi H, Nourian S. Automatic assessment of mitral regurgitation severity based on extensive textural features on 2D echocardiography videos. Comput Biol Med. 2016;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDouedi S, Douedi H. Mitral Regurgitation [Internet]. Statpearls. 2024 [cited 2024 Nov 29]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/books/NBK553135/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/books/NBK553135/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGandhi S, Mosleh W, Shen J, Chow CM. Automation, machine learning, and artificial intelligence in echocardiography: A brave new world. Vol. 35, Echocardiography. 2018.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiller GP, Kempny A, Babu-Narayan S V., Henrichs M, Brida M, Uebing A, et al. Machine learning algorithms estimating prognosis and guiding therapy in adult congenital heart disease: Data from a single tertiary centre including 10 019 patients. Eur Heart J. 2019;40(13).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiastuti LD, Nursakina Y. Diagnostic accuracy of artificial intelligence models in detecting congenital heart disease in the second-trimester fetus through prenatal cardiac screening: a systematic review and meta-analysis. Front Cardiovasc Med [Internet]. 2025 [cited 2025 Apr 29];12:1473544. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC11891181/\u003c/span\u003e\u003cspan address=\"https://pmc.ncbi.nlm.nih.gov/articles/PMC11891181/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoodbhoy Z, Jiwani U, Sattar S, Salam R, Hasan B, Das JK. Diagnostic Accuracy of Machine Learning Models to Identify Congenital Heart Disease: A Meta-Analysis. Vol. 4, Frontiers in Artificial Intelligence. 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhorbani A, Ouyang D, Abid A, He B, Chen JH, Harrington RA, et al. Deep learning interpretation of echocardiograms. NPJ Digit Med. 2020;3(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTromp J, Bauer D, Claggett BL, Frost M, Iversen MB, Prasad N, et al. A formal validation of a deep learning-based automated workflow for the interpretation of the echocardiogram. Nat Commun. 2022;13(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNazar W, Nazar K, Daniłowicz-Szymanowicz L. Machine Learning and Deep Learning Methods for Fast and Accurate Assessment of Transthoracic Echocardiogram Image Quality. Life [Internet]. 2024 Jun 1 [cited 2025 Apr 29];14(6):761. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.mdpi.com/\u003c/span\u003e\u003cspan address=\"https://www.mdpi.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e2075-1729/14/6/761/htm\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHirata Y, Kusunose K. AI in Echocardiography: State-of-the-art Automated Measurement Techniques and Clinical Applications. JMA J [Internet]. 2024 [cited 2025 Apr 29];8(1):141. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC11799715/\u003c/span\u003e\u003cspan address=\"https://pmc.ncbi.nlm.nih.gov/articles/PMC11799715/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang L, Zuo HJ, Chen C. Artificial intelligence in echocardiography: Applications and future directions. Fundamental Research [Internet]. 2025 Apr 25 [cited 2025 Apr 29]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://linkinghub.elsevier.com/retrieve/pii/S2667325825001025\u003c/span\u003e\u003cspan address=\"https://linkinghub.elsevier.com/retrieve/pii/S2667325825001025\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKusunose K. Transforming Echocardiography: The Role of Artificial Intelligence in Enhancing Diagnostic Accuracy and Accessibility. Internal Medicine [Internet]. 2024 Feb 1 [cited 2025 Apr 29];64(3):331. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC11867741/\u003c/span\u003e\u003cspan address=\"https://pmc.ncbi.nlm.nih.gov/articles/PMC11867741/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlabdaljabar MS, Hasan B, Noseworthy PA, Maalouf JF, Ammash NM, Hashmi SK. Machine Learning in Cardiology: A Potential Real-World Solution in Low- and Middle-Income Countries. J Multidiscip Healthc [Internet]. 2023 [cited 2025 Apr 29];16:285\u0026ndash;95. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.tandfonline.com/doi/pdf/\u003c/span\u003e\u003cspan address=\"https://www.tandfonline.com/doi/pdf/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/JMDH.S383810\u003c/span\u003e\u003cspan address=\"10.2147/JMDH.S383810\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe J, Baxter SL, Xu J, Xu J, Zhou X, Zhang K. The practical implementation of artificial intelligence technologies in medicine. Vol. 25, Nature Medicine. 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDey D, Slomka PJ, Leeson P, Comaniciu D, Shrestha S, Sengupta PP, et al. Artificial Intelligence in Cardiovascular Imaging: JACC State-of-the-Art Review. Vol. 73, Journal of the American College of Cardiology. 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhmed MI, Spooner B, Isherwood J, Lane M, Orrock E, Dennison A. A Systematic Review of the Barriers to the Implementation of Artificial Intelligence in Healthcare. Cureus. 2023;\u003c/span\u003e\u003c/li\u003e \u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"artificial intelligence, deep learning, echocardiography, mitral regurgitation, screening","lastPublishedDoi":"10.21203/rs.3.rs-6626740/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6626740/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eMitral regurgitation (MR) is a prevalent and potentially progressive cardiovascular condition, necessitating early detection to facilitate timely intervention and optimize patient outcomes. The increasing demand for efficient and precise diagnostic strategies has underscored the potential of artificial intelligence (AI) in clinical practice. By leveraging advanced AI algorithms, automated MR screening has the capacity to enhance the detection and classification of disease severity, thereby assisting clinicians in making well-informed decisions. This study aims to evaluate the accuracy and efficacy of AI-based echocardiographic analysis in the early diagnosis of MR.\u003c/p\u003e\u003ch2\u003eMain Text\u003c/h2\u003e \u003cp\u003eA comprehensive literature search was conducted across five databases, PubMed, Scopus, ScienceDirect, ProQuest, and Cochrane. Studies employing AI algorithms to analyze echocardiographic images for MR detection and severity classification were included. The methodological quality of each study was assessed using the QUADAS-2 tool for diagnostic accuracy studies. A quantitative meta-analysis was performed utilizing Meta-DiSc with a random-effects model. In total, nine studies met the inclusion criteria. Utilization of AI in echocardiographic detection of MR yield a pooled sensitivity of 0.85 (95% CI: 0.86\u0026ndash;0.86), specificity of 0.83 (95% CI: 0.82\u0026ndash;0.83), and an area under the curve (AUC) of 0.9745. Accurate detection and severity classification of MR could significantly improve the efficiency of treatment strategies. By reducing reliance on specialized personnel and enabling the use of portable imaging devices, AI can lower operational costs and expand access to high-quality diagnostics. Furthermore, AI integration has the potential to streamline clinical workflows, decrease diagnostic delays, and optimize resource allocation. However, successful implementation requires addressing challenges related to model generalizability, regulatory standards, clinician training, and integration into existing healthcare systems.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eIn conclusion, AI-assisted echocardiographic analysis presents a promising advancement in MR diagnostics, with the potential to enhance healthcare accessibility, particularly in resource-limited settings.\u003c/p\u003e","manuscriptTitle":"Integrating Artificial Intelligence for Mitral Regurgitation Assessment: A systematic review and meta-analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-20 06:46:22","doi":"10.21203/rs.3.rs-6626740/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"316b5ee5-6aeb-4436-bcb4-a2b7e528fe70","owner":[],"postedDate":"May 20th, 2025","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-16T21:19:05+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-15T20:24:25+00:00","index":45,"fulltext":""},{"type":"reviewerAgreed","content":"237635117618584190990033715396480427203","date":"2026-05-15T20:12:30+00:00","index":44,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-16T21:23:54+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-20 06:46:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6626740","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6626740","identity":"rs-6626740","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}