Background: Heart failure with preserved ejection fraction (HFpEF) is a heterogeneous syndrome characterized by frequent underdiagnosis, diverse etiologies, and limited therapeutic options. Given its complexity, artificial intelligence (AI) and machine learning (ML) offer promising avenues to decode high-dimensional, multi-modal healthcare data. This review aims to synthesize the current landscape of AI/ML applications in HFpEF, evaluating their potential to address critical unmet clinical needs. Methods: We conducted a comprehensive review of the literature focusing on AI/ML paradigms in HFpEF. Key methodological frameworks were examined, including supervised, unsupervised, semi-supervised, and reinforcement learning, alongside advanced techniques such as deep learning and natural language processing (NLP). The analysis focused on the application of these techniques across four domains: diagnosis, sub-phenotyping, risk prediction, and optimization of diagnostic modalities, with specific emphasis on studies incorporating external validation. Results: Current evidence demonstrates that AI approaches effectively enhance diagnostic accuracy and facilitate the identification of distinct HFpEF phenotypes beyond traditional classifications. These technologies show significant utility in refining prognostic assessments and optimizing diagnostic testing strategies. Furthermore, ML-driven analytics provide a robust framework for improving patient selection and streamlining clinical trial design, potentially overcoming historical barriers to drug development in this population. Conclusions: AI represents a transformative tool capable of dissecting the heterogeneity of HFpEF to enable precision medicine. While the potential to improve clinical outcomes is substantial, challenges regarding model interpretability, bias, and clinical integration persist. Future efforts must focus on rigorous external validation and prospective trials to ensure the responsible translation of these technologies into routine clinical practice.

Left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) are essential for the diagnosis, clinical decision-making, and prognosis of cardiovascular disease. However, accurate assessments of LVEF and GLS by echocardiography are hampered by inter-observer variability, time-consuming, and labor-intensive. This study aimed to develop an automated method to accurately and rapidly assess LVEF and GLS. Based on the datasets of 500 patients (1,500 videos) from the internal center and 363 patients (1,089 videos) from four external centers, we successfully developed a dual-flow convolutional neural network called Echo-DFCNN, which allowed for synchronous acquisition of LVEF and GLS. We evaluated the performance of the Echo-DFCNN in a cardiac magnetic resonance (CMR) validation dataset composed of 67 patients. On the internal test dataset, the AI and manual measurements of LVEF demonstrated a median absolute error of 3.02% and a mean absolute error of 3.94%. AI-predicted LVEF showed good agreement with manually measured LVEF, with an ICC of 0.927, a bias of 0.89%, and a LOA of -10.91 to 12.69. For GLS, the median absolute error and mean absolute error between AI and manual measurements were 1.43% and 1.83%. AI-predicted GLS exhibited high agreement with manually measured GLS (ICC = 0.913; bias = -1.22%, LOA = -5.12 to 2.68). In addition, Echo-DFCNN maintained good performance when applied to external validation datasets. In the CMR validation dataset, the AI model showed good agreement with CMR measurements for both LVEF and GLS. Echo-DFCNN achieves simultaneous and precise assessment of LVEF and GLS in the study cohorts, demonstrating its potential for robust performance across a wide range of cardiac functions, different image qualities, and machine types.

Background: Papillary muscles (PMs) are important for mitral valve competence and left ventricular mechanics, but accurate evaluation is often limited by poor visualization in conventional echocardiographic views. We developed papillary muscle-targeted (PM-targeted) echocardiographic views to improve PM visualization and aimed to validate this approach and establish normative reference values in healthy adults. Methods: In protocol 1, posteromedial papillary muscle (PPM) length and maximum diameter measured using PM-targeted and standard views were compared with anatomic measurements in ten ex vivo porcine hearts. In protocol 2, measurements of the anterolateral papillary muscle (APM) and PPM were compared between PM-targeted and standard views in 100 healthy adults. In protocol 3, PM structural, spatial, and functional parameters were measured using PM-targeted views in 245 healthy adults. In protocol 4, PM measurements obtained from 2D PM-targeted views were compared with 3D echocardiographic measurements in 50 patients with ventricular functional mitral regurgitation (VFMR); PM parameters in VFMR were also compared with those in healthy adults. Results: In protocol 1, PM-targeted views showed stronger correlation with anatomic measurements for PPM length than standard views (0.966 vs. 0.752, p = 0.049), while standard views underestimated PPM length. In protocol 2, PM-targeted views enabled complete visualization of APM and PPM and yielded longer PM lengths and smaller maximum diameters than standard views. In protocol 3, males had larger PM maximum diameters and longer tip-to-annulus distances than females (all p < 0.05). With aging, interpapillary distance reduction (ΔIPMD), IPMD fractional shortening (IPMD-FS), and APM length decreased, whereas end-systolic IPMD increased (all p < 0.05). PM parameters correlated positively with body surface area (all p < 0.05). In protocol 4, PM measurements obtained from 2D PM-targeted views showed no differences from 3D echocardiographic measurements and demonstrated good correlation and agreement across assessed PM parameters against 3D echocardiographic measurement as a standard reference. Compared with healthy adults, patients with VFMR showed altered PM geometry/remodeling patterns. Conclusions: PM-targeted echocardiographic views improve visualization and measurement of papillary muscles and provide normative reference values, facilitating more accurate evaluation of PM-related abnormalities in clinical practice.

Background Right ventricular free wall longitudinal strain (RVFWLS) assessed by three-dimensional speckle-tracking echocardiography (3D-STE), two-dimensional (2D) STE and cardiac magnetic resonance feature tracking (CMR-FT) has been reported to correlate with the degree of right ventricular myocardial fibrosis (RVMF), but the value of 3D-STE in evaluating RVMF compared with CMR-FT has not been investigated in patients with advanced heart failure. Aims This study aimed to determine which RVFWLS obtained by 2D-STE, 3D-STE, and CMR-FT, is the most robust noninvasive imaging marker for evaluating RVMF in patients with advanced heart failure using histopathological RVMF as the gold standard. Methods 151 patients (118 male; mean age, 40±18 years) with advanced heart failure who underwent CMR, 2D and 3D echocardiographic examinations before heart transplantation were recruited in this study. RVFWLS was obtained from 2D-STE, 3D-STE and CMR-FT. The degree of RVMF was evaluated using late gadolinium enhancement (LGE) of CMR and Masson’s staining in right ventricular myocardial samples, and patients were divided into three groups according to tertiles of histopathological RVMF. Results Patients with severe RVMF had lower 3D-RVFWLS, 2D-RVFWLS and CMR-derived RVFWLS, and higher CMR-LGE compared with those with mild and moderate RVMF. RVMF was strongly correlated with CMR-derived RVFWLS (r = 0.73, p < 0.001), 3D-RVFWLS (r = 0.72, p < 0.001) and CMR-LGE (r = 0.71, p < 0.001), and moderately correlated with 2D-RVFWLS (r = 0.56, p < 0.001). The correlations of CMR-derived RVFWLS and 3D-RVFWLS with RVMF were not different from correlation of CMR-LGE with RVMF (p > 0.05). CMR-LGE, CMR-derived RVFWLS and 3D-RVFWLS had similar performance in identifying severe RVMF (AUC: 0.90 vs 0.87 vs 0.85, all p > 0.05). The model with CMR-derived RVFWLS (R2 = 0.515, p < 0.001) had a similar ability to assess RVMF as the model with 3D-RVFWLS (R2 = 0.499, p < 0.001), but better than the model with 2D-RVFWLS (R2 = 0.340, p < 0.001). Conclusions 3D-RVFWLS and CMR-derived RVFWLS are strongly correlated with RVMF in patients with advanced heart failure. 3D-STE may be a promising technique that correlates with RVMF, providing a similar accuracy as CMR-FT in assessing RVMF.

#These authors contributed equally.

BACKGROUND:Right ventricular free wall longitudinal strain (RVFWLS) is a sensitive marker of RV dysfunction after heart transplantation (HT), and automated RVFWLS may improve efficiency. This study evaluated the accuracy and prognostic value of automated echocardiographic RVFWLS using cardiac magnetic resonance imaging (CMR) as reference. METHODS:A total of 150 HT recipients undergoing echocardiography and CMR within 3 days were retrospectively analysed. The accuracy and prognostic value of fully and semiautomated RVFWLS were compared with CMR. Image quality was graded as "optimal" or "acceptable" to assess its influence on automated measurements. RESULTS:Both fully and semiautomated methods correlated with CMR (r = 0.727 and 0.863; P < 0.001), with the semiautomated approach showing smaller bias, narrower limits of agreement, and lower coefficient of variation. The subgroup of "acceptable" image quality reduced the accuracy of automated RVFWLS. During a median 37-month follow-up, 29 patients experienced adverse events. In multivariable Cox analysis, semiautomated RVFWLS (hazard ratio [HR] = 1.499; AIC = 229; C-index = 0.807) showed prognostic performance similar to CMR (HR = 1.570; AIC = 216; C-index = 0.852) and outperformed fully automated RVFWLS (HR = 1.284; AIC = 249; C-index = 0.734). Receiver operating characteristic analysis confirmed the superiority of semiautomated RVFWLS over the fully automated method in predicting adverse events (area under the receiver operating characteristic curve 0.845 vs 0.735; P = 0.002). CONCLUSIONS:Automated RVFWLS showed good accuracy and prognostic value in HT patients as validated by CMR. The semiautomated approach may be preferable for post-HT follow-up owing to its superior performance, whereas the fully automated method may be a potentially acceptable alternative when image quality is adequate.

Background: Three-dimensional (3D) speckle-tracking echocardiography (STE) and cardiovascular magnetic resonance (CMR) feature tracking (FT) have been reported to correlate with the extent of left ventricular myocardial fibrosis (LVMF), but the value of 3D-STE in predicting LVMF compared with CMR-FT has not been investigated in patients with end-stage heart failure (HF). Methods: A total of 155 patients who underwent CMR, two-dimensional (2D), and 3D echocardiographic examinations before heart transplantation were enrolled. Left ventricular global radial strain (GRS), global circumferential strain (GCS), and global longitudinal strain (GLS) were obtained from CMR-FT, 3D-STE, and 2D-STE. The degree of LVMF was quantified using Masson’s trichrome staining in left ventricular myocardial samples from the explanted hearts, and patients were divided into three groups according to tertiles of histologic LVMF. Results: 3D-GLS, 2D-GLS, CMR-GLS, and CMR-GCS were lower in patients with severe LVMF than in those with mild and moderate LVMF. LVMF was strongly correlated with CMR-GLS and 3D-GLS (r = 0.74, 0.73, respectively; both P < 0.001), and moderately correlated with 2D-GLS (r = 0.63, P < 0.001). CMR-GLS and 3D-GLS demonstrated similar diagnostic performance in identifying severe LVMF (area under the curve: 0.88 vs. 0.86, P > 0.05). The model with CMR-GLS (R2 = 0.550, P < 0.001) had a similar predictive value for the degree of LVMF as the model with 3D-GLS (R2 = 0.528, P < 0.001), and outperformed the model with 2D-GLS (R2 = 0.383, P < 0.001). Conclusion: Both CMR-GLS and 3D-GLS are strongly correlated with LVMF, and are promising non-invasive imaging parameters for the assessment of LVMF in patients with end-stage HF.

Abstract Background Echocardiographic Myocardial work (MW) has potential value in hypertrophic obstructive cardiomyopathy (HOCM). This study aimed to utilize corrected MW indices to characterize left ventricular (LV) myocardial mechanical remodeling and evaluate the extent of myocardial fibrosis (MF) in patients with HOCM. Methods We prospectively studied 41 patients with HOCM undergoing septal myectomy (SM). 21 patients underwent intraoperative invasive pressure measurement to validate the noninvasive left ventricular systolic pressure (LVSP) estimation and corrected MW analysis methods. Transthoracic echocardiography was performed in all patients at baseline and 3–6 months after SM. Preoperative and postoperative parameters such as global work index (GWI), global constructive work (GCW), global wasted work (GWW), and global work efficiency (GWE) were analyzed to investigate the characteristics of LV myocardial mechanical remodeling. The degree of histological MF was evaluated to determine the correlation between corrected MW parameters and MF. Results Noninvasive LVSP estimated by adding systolic blood pressure to the peak LV outflow tract gradient was well consistent with invasively measured LVSP (r = 0.98, P < 0.001; ICC = 0.96, P < 0.001). After SM, GWI, GCW, and GWE were significantly decreased (all P < 0.001), and GWW was significantly increased in HOCM patients (P = 0.002). Postoperatively, all patients exhibited new-onset complete left bundle branch block. Corrected GWI (R²=0.22, P = 0.002) and GCW (R²=0.25, P < 0.001) were independently associated with the extent of MF. Conclusion We validated a corrected method for analyzing MW in HOCM patients. HOCM patients may experience reduced metabolism and compromised contraction coordination after SM. GWI and GCW are associated with the level of MF.

The echocardiography is the first-line imaging modality in detecting the cardiac lipoma. Contrast-enhanced echocardiography improves its structural definition and characteristics of blood supply to exclude thrombus and malignant tumors. We introduced a case that large cardiac mass involving nearly the whole left ventricular cavity and papillary muscles without any complications. Multimodal imaging has confirmed lipoma before surgery. However, rather than recommending conservative treatment in accordance with guidelines, surgical intervention was performed to prevent future hemodynamic abnormalities. Combined with multimodal imaging, we showed a rare case on comprehensive evaluation of left ventricular silent lipoma and provided new clues for surgical strategy, which were different from guideline recommendations.

Background:Tricuspid valve (TV) replacement with surgical bioprosthetic, surgical mechanical, or transcatheter prostheses is a critical intervention for severe tricuspid regurgitation or stenosis. However, comprehensive echocardiographic comparisons of hemodynamic performance across these prosthesis types remain limited, especially for novel transcatheter prostheses. This study aimed to systematically analyze the transthoracic echocardiographic (TTE) Doppler parameters of three types of tricuspid prostheses with normal function to provide accurate references for the management of patients following clinical TV replacement (TVR). Methods:This retrospective study included 62 patients with surgical bioprosthetic valves, 52 with surgical mechanical valves, and 25 with transcatheter valves. Clinical data, along with comprehensive two-dimensional (2D) and Doppler echocardiographic parameters, were collected from normal tricuspid prostheses. Doppler parameters were measured at the TV position, including peak early tricuspid diastolic velocity (E velocity), mean gradient (MGTV), velocity-time integral (VTITV), pressure half-time (PHTTV), the ratio of VTITV to the VTI of the left ventricular outflow tract (VTI ratio), the effective orifice area (EOATV), and the indexed effective orifice area (IEOATV). Results:The normal ranges (95% confidence interval) of transthoracic Doppler echocardiographic parameters for the three types of tricuspid prostheses were established for surgical bioprosthetic valves (E velocity, 1.40-1.59 m/s; MGTV, 3.90-4.93 mmHg; VTITV, 41.8-47.2 cm; PHTTV, 135.8-157.8 ms; VTI ratio, 2.03-2.31; EOATV, 1.44-1.69 cm2; IEOATV, 0.92-1.09 cm2/m2), surgical mechanical valves (E velocity, 1.40-1.58 m/s; MGTV, 3.08-3.93 mmHg; VTITV, 34.6-39.9 cm; PHTTV, 106.7-122.8 ms; VTI ratio, 1.70-2.04; EOATV, 1.73-2.10 cm2; IEOATV, 1.08-1.31 cm2/m2), and transcatheter valves (E velocity, 1.21-1.51 m/s; MGTV, 2.58-4.83 mmHg; VTITV, 32.7-40.4 cm; PHTTV, 107.5-132.2 ms; VTI ratio, 1.60-1.99; EOATV, 1.80-2.44 cm2; IEOATV, 1.15-1.54 cm2/m2). Conclusions:This study established prosthesis-specific echocardiographic reference values for normal tricuspid prostheses. Values outside these ranges may indicate dysfunction, although the clinical correlation remains to be determined. These findings can enhance the postoperative monitoring of tricuspid prostheses.

BACKGROUND:The impaired right ventricular (RV) function may occur after heart transplantation (HT). This study aimed to develop a straightforward RV function score (RVFS) using echocardiography to predict adverse clinical events in a large cohort of subjects after HT. METHODS:A total of 357 consecutive patients post HT who underwent echocardiography at our single institution were retrospectively included. RV systolic function was evaluated by echocardiography. A multivariate Cox regression analysis was performed to identify the independent predictors of RV function for adverse clinical events. RESULTS:During a median follow-up of 39 months from the date of the echocardiography, a composite of adverse events that included death and major adverse cardiac events occurred in 51 patients. The multivariate Cox analysis revealed that RV fractional area change, tricuspid annular plane systolic excursion, and RV free wall longitudinal strain were independent predictors of outcomes after HT. The RVFS was constructed based on these 3 parameters, assigning a value of 1 when a predictor was below its cutoff value from receiver operating characteristic curves and 0 when above. The RVFS outperformed separate RV function parameters in predicting outcomes (area under the curves: 0.84 versus 0.64-0.78, P<0.05). Moreover, incorporating the RVFS into the base clinical model significantly improved the C-statistic of the prediction model (C-statistic from 0.70 to 0.83, P<0.001). CONCLUSIONS:The RVFS, a readily obtainable echo score that combines multiple RV function parameters, exhibits incremental value and significant potential as a robust predictor for adverse outcomes in clinically stable patients after HT.

Background Accurate and early identification of impaired left ventricular (LV) function is essential to the optimal timing of intervention for primary mitral regurgitation. Myocardial work derived from noninvasive pressure‐strain loop is a novel and promising afterload‐independent approach to evaluate LV performance. We hypothesized that it may provide diagnostic and prognostic utility in these patients. This study aimed to evaluate myocardial work parameters in patients with significant primary mitral regurgitation, and explore their association with postinterventional LV ejection fraction and clinical composite events. Methods The study prospectively enrolled 180 patients with severe primary mitral regurgitation at baseline and patients were followed up for postinterventional LV function (>12 months) and clinical events (24.0 [23.3–24.6] months). Logistic regression and Cox proportional hazards regression analyses were performed as appropriate. Results Compared with patients exhibiting postinterventional LV ejection fraction ≥50%, individuals with LV ejection fraction <50% demonstrated lower LV global longitudinal strain, global work index (GWI), global constructive work (GCW), and global work efficiency (P<0.001 for all) and higher global wasted work (P=0.001). Preinterventional LV global longitudinal strain, GWI, GCW, and global work efficiency were independent predictors of postinterventional LV dysfunction (P<0.05 for all). The predictive power of logistic regression models comprising LV global longitudinal strain, GWI, and GCW were similar but stronger than the model comprising LV ejection fraction. Preinterventional LV global longitudinal strain, GWI, GCW, and global work efficiency were independently associated with the risk of composite clinical events during follow‐up (P<0.001 for all). Conclusions Myocardial work parameters, especially GWI, GCW, and global work efficiency, are independent predictors of postinterventional LV dysfunction and are associated with the occurrence of postinterventional clinical composite events.