Abstract
Background
Sustained ventricular tachycardia (VT) is a major cause of sudden cardiac death in structural heart disease (SHD). Risk stratification based on left ventricular ejection fraction (LVEF) remains suboptimal. This study aimed to develop an exploratory model integrating late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) scar characteristics with computational VT simulation to discriminate sustained VT, compared against LVEF.
Methods
In this single-centre retrospective study, consecutive patients with suspected SHD who underwent LGE-CMR were enrolled. Patients with LGE-positive underwent quantitative scar analysis, personalized digital heart-twin reconstruction, and virtual electrophysiological simulation.
Results
Among 82 patients (mean age 59.1 years; 82.9% male), 15 experienced sustained VT. Lower LVEF was associated with sustained VT (P = 0.008; C-statistic: 0.730), as was visual LGE positivity (P = 0.023; C-statistic: 0.634). Among LGE-positive patients, those with sustained VT had a larger core scar [10.0 [5.9, 14.8]% vs. 2.2 [1.1, 4.7]%; P < 0.001], larger grey zone [14.0 [10.9, 14.7]% vs. 5.7 [2.3, 9.2]%; P < 0.001], higher VT inducibility [50.0 [22.9, 66.9]% vs. 0 (0, 21.1)%; P < 0.001], and a greater number of inducible VT circuits [2 [2, 4] vs. 0 [0, 2]; P < 0.001]. An integrated LGE-VTsim index, derived from these four parameters, effectively discriminated sustained VT (C-statistic: 0.816) and remained independently associated with sustained VT after adjusting for LVEF (OR: 1.029; 95% CI: 1.008–1.051; P = 0.007).
Conclusion
The LGE-VTsim index, integrating quantitative LGE-CMR scar assessment with computational simulation, represents a promising exploratory parameter in identifying patients at risk of sustained VT in SHD, beyond LVEF.
Keywords: cardiac magnetic resonance, computational simulation, risk stratification, structural heart disease, ventricular tachycardia
1. Introduction
Sustained ventricular tachycardia (VT) is a major cause of sudden cardiac death (SCD) in patients with structural heart disease (SHD). Contemporary pharmaceutical management has substantially reduced the incidence of SCD in patients with heart failure. However, a persistent residual risk remains despite guideline-directed medical therapy. Implantable cardioverter–defibrillator (ICD) implantation has therefore been associated with a further reduction in SCD risk, underscoring the continued necessity of ICD therapy and highlighting the urgent need for more refined risk stratification tools (1). Current guidelines primarily rely on a left ventricular ejection fraction (LVEF) ≤ 35% to guide primary prevention ICD therapy (2). Nevertheless, a substantial proportion of SCD events occur in patients with SHD and LVEF >35% (3, 4), underscoring the limitations of LVEF as a standalone risk marker.
Late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) enables detailed characterisation of the arrhythmogenic substrate, including delineation of the infarct core and the border zone (grey zone) that are critical for re-entrant VT circuits (5–7) Although LGE-based scar quantification predicts arrhythmic risk in both ischaemic (ICM) and non-ischaemic cardiomyopathies (NICM) (8, 9), image-derived parameters alone may not fully capture the dynamic electrophysiological properties that are critical for VT inducibility and maintenance.
Recent advances in computational modelling have enabled the creation of patient-specific “digital twins” that integrate LGE-CMR–derived scar morphology with simulated electrophysiology to non-invasively assess VT inducibility (10–12). This approach seeks to capture the complex interplay between scar architecture and arrhythmia vulnerability.
In this study, we retrospectively analysed patients with SHD undergoing LGE-CMR to address four questions: 1) whether LVEF ≤35% represents the optimal cut-off for identifying sustained VT; 2) whether visual LGE assessment outperforms LVEF; 3) whether a model combining quantitative LGE characteristics with computational VT simulation surpasses LVEF alone; and 4) whether such a model can differentiate sustained from non-sustained ventricular arrhythmias (VAs). Our objective was to develop an integrated VT identification tool and compared with LVEF-based approaches.
2. Materials and methods
2.1. Study design and population
This single-centre retrospective study was conducted at Beijing Chaoyang Hospital and consecutively enrolled patients with SHD who underwent LGE-CMR between September 2024 and February 2025. The study protocol was approved by the Institutional Review Board of Beijing Chaoyang Hospital (2025-KE-921), and all participants provided informed consent.
Eligible patients met at least one of the following criteria: (1) LVEF <50%; (2) prior myocardial infarction, including ST-segment elevation myocardial infarction (STEMI) or non-ST-segment elevation myocardial infarction (NSTEMI); (3) diagnosed cardiomyopathy, including dilated cardiomyopathy (DCM) or hypertrophic cardiomyopathy (HCM); (4) echocardiographic evidence of regional wall motion abnormalities; (5) electrocardiographic documentation of VA; or (6) planned implantation of an ICD or cardiac resynchronisation therapy defibrillator. Exclusion criteria were: (1) arrhythmogenic right ventricular cardiomyopathy; (2) inherited channelopathies; (3) age <18 years; (4) estimated glomerular filtration rate <30 mL/min; (5) contraindications to CMR; or (6) poor LGE-CMR image quality. Genetic cardiomyopathies were not systematically assessed owing to limited availability of genetic testing.
2.2. Clinical data collection
SHD was classified as ICM or NICM. ICM was diagnosed based on any of the following: a history of ischaemic heart disease or coronary revascularisation; coronary artery stenosis ≥50%; inducible ischaemia; or an LGE pattern consistent with myocardial infarction (13). NICM was defined by a non-coronary LGE distribution in the absence of significant coronary artery stenosis (14, 15). Data on LVEF, New York Heart Association (NYHA) functional class categories, comorbidities, and medications were systematically collected.
All patients underwent clinical and arrhythmia assessments within a 6-month peri-CMR period, including outpatient consultations, electrocardiography, ICD interrogations, and 24-hour Holter monitoring. Ventricular arrhythmias (VAs) were categorised as: (1) sustained VT (>30 s or requiring intervention due to haemodynamic compromise, including appropriate ICD therapy); (2) non-sustained VA [frequent premature ventricular contractions (PVCs) ≥ 1,000 per 24 h, with or without non-sustained VT] (16, 17); or (3) no clinically relevant VA (<1,000 PVCs per 24 h without sustained VT).
2.3. Study group definitions
Patients were classified into Event and Event-free groups according to the occurrence of sustained VT during the period from one week before to 6 months after the LGE-CMR examination. The Event group included patients with appropriate ICD therapy, SCD, or documented sustained VT, whereas all remaining patients comprised the Event-free group.
2.4. Image acquisition, digital twin reconstruction, and electrophysiological simulation
All participants underwent LGE-CMR using 3.0-T scanners (Siemens Prisma or Vida). Epicardial and endocardial contours were automatically segmented and manually refined to exclude the blood pool and epicardial fat. Myocardial tissue was then classified into non-scar and scar regions, with scar further subdivided into core scar (pixels >50% of the signal intensity range) and grey zone according to pixel intensity thresholds. Segmented images were interpolated to a 0.4 mm resolution using CardioViz3D. Scar geometry was reconstructed using the log-odds method and integrated into patient-specific biventricular finite-element meshes (approximately 400 μm element size), in which myocardial fibre orientation was assigned from −40° at the epicardium to +65° at the endocardium using a rule-based approach.
The electrophysiological properties of non-scared myocardium, grey zone, and core scar were simulated using the ten Tusscher model. Ionic current conductances in the grey zone were scaled to 30% (IKr), 20% (IKs), 31% (ICaL), and 38% (INa) of normal values, resulting in conduction velocities of 0.60 m/s in healthy myocardium and 0.30 m/s in the grey zone, while the core scar was treated as electrically passive. VT inducibility was assessed in the openCARP environment by applying a standardised rapid pacing protocol, consisting of a six-beat S1 train (600 ms) followed by up to three premature stimuli (S2–S4), at eight ventricular sites defined by the American Heart Association segmented model (18). Induced re-entrant VT was simulated for 10 s to confirm stability, and circuit locations were mapped for subsequent analysis. The LGE segmentation and virtual electrophysiological study pipeline was implemented in a largely automated manner to minimise operator-dependent variability. Throughout the entire process, the engineer performing the simulations was blinded to all patient outcomes to prevent bias.
2.5. Statistical analysis
Continuous variables are presented as mean ± standard deviation or median with interquartile range, and categorical variables as frequencies and percentages. Between-group comparisons were performed using the Chi-square test (χ2 test) or Fisher's exact test, as appropriate. Correlation analyses were performed to assess collinearity and variable interdependence, followed by exploratory factor analysis (EFA) to extract latent factors and reduce data dimensionality for subsequent modeling (LGE-VTsim index). Prior to EFA, all variables were standardized to z-scores (mean = 0, SD = 1) to eliminate scale differences and ensure equal contribution to the extracted factor. The discriminatory performance of LVEF, visual LGE assessment, and the LGE-VTsim index for sustained VT was evaluated using the C-statistic, with comparisons conducted using DeLong's test. Optimal cut-off values were determined by maximising the Youden index. Logistic regression was used to assess associations with outcomes, which are expressed as odds ratios (OR) with 95% confidence intervals (CI). The incremental discriminative value of the LGE-VTsim index over LVEF was assessed using the net reclassification index (NRI), integrated discrimination improvement (IDI), and decision-curve analysis (DCA). To address the risk of overfitting, optimism-corrected C-statistics were calculated via bootstrap resampling. Calibration of the model incorporating the LGE-VTsim index was assessed by comparing predicted probabilities with observed outcomes using a calibration plot with loess smoothing; the 45° line represented perfect calibration. Statistical analyses were performed using SPSS (version 31.0) and R (version 4.4.2). A two-sided P < 0.05 was considered statistically significant.
3. Results
3.1. Baseline characteristics and endpoints
A total of 97 patients undergoing LGE-CMR were retrospectively enrolled. After exclusion of 15 patients (2 with arrhythmogenic right ventricular cardiomyopathy, 3 due to poor LGE-CMR image quality, and 10 with incomplete arrhythmia assessment during the 6-month peri-CMR screening window), 82 patients constituted the final study population (Figure 1). The mean age was 62.0 (53.0, 68.0) years, and the majority of patients were male (82.9%). The mean LVEF was 45.4% ± 14.4%; 20 patients (24.4%) had an LVEF ≤35%, and 64 (78.0%) were visually assessed as LGE-positive. Detailed clinical characteristics are presented in Table 1.
Figure 1.
Study flow chart and clinical outcomes. The flowchart depicts the selection of patients with suspected SHD based on LGE-CMR findings and their subsequent classification by 6-month arrhythmic outcomes. LGE-CMR, late gadolinium enhancement cardiac magnetic resonance; VT, ventricular tachycardia; SCD, sudden cardiac death; ICD, implantable cardioverter-defibrillator; VA, ventricular arrhythmia; PVCs, premature ventricular contractions.
Table 1.
Baseline characteristics of the study population.
| Characteristic | Event (n = 15) | Event-free (n = 67) | P |
|---|---|---|---|
| Female | 1 (6.7%) | 13 (19.4%) | 0.421 |
| Age (years) | 56 (53.5, 68.5) | 62 (54, 67) | 0.723 |
| Body surface area (m2) | 1.91 ± 0.12 | 1.86 ± 0.18 | 0.370 |
| Ventricular arrhythmia Burden % | 0.3 (0.2, 0.95) | 0.3 (0.1, 1.15) | 0.516 |
| Diabetes | 4 (30.8%) | 20 (29.9%) | 0.947 |
| Hypertension | 5 (38.5%) | 37 (55.2%) | 0.268 |
| Ischemic cardiomyopathy | 9 (60.0%) | 44 (65.7%) | 0.678 |
| NYHA function class | |||
| I | 1 (6.7%) | 14 (20.9%) | 0.425 |
| II | 8 (53.3%) | 35 (52.2%) | |
| III | 5 (33.3%) | 16 (23.9%) | |
| IV | 1 (6.7%) | 2 (3.0%) | |
| Medications | |||
| β-blocker | 15 (100%) | 51 (76.1%) | 0.035 |
| ACEI/ARB | 15 (100%) | 49 (73.1%) | 0.023 |
| Aldosterone antagonist | 14 (93.3%) | 24 (35.8%) | <0.001 |
| Loop diuretic | 5 (33.3%) | 11 (16.4%) | 0.135 |
| LGE-CMR | |||
| LVEF % | 35.6 ± 13.3 | 47.4 ± 13.9 | 0.008 |
| LVEF ≤35% | 7 (50.0%) | 13 (19.4%) | 0.016 |
| LVEF>50% | 1 (7.1%) | 30 (44.8%) | 0.008 |
| LV mass (g) | 153.8 (128.6, 174.1) | 140.2 (105.0, 171.4) | 0.450 |
| RVEF (%) | 50.3 ± 16.7 | 52.9 ± 10.3 | 0.534 |
| LGE-positive | 15 (100%) | 49 (73.1%) | 0.023 |
ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; LV, left ventricular; LVEF, left ventricular ejection fraction; LGE, late gadolinium enhancement; NYHA, New York Heart Association; RVEF, right ventricular ejection fraction.
Fifteen patients experienced clinical sustained VT events (Event Group). Of these, 12 had VT episodes shortly before CMR, and 3 developed newly documented sustained VT after CMR, all of whom received appropriate ICD therapy. All patients in the Event Group were receiving β-blocker therapy. For clinical management, 8 patients received an ICD, 5 underwent VT ablation (including 3 with concurrent ICD implantation), and 5 declined invasive intervention (Figure 1).
3.2. Scar quantification and computational VT simulation in LGE-positive patients
Analyses focused on the 64 LGE-positive patients [age 62.0 (53.5, 69.0) years; 84.4% male], including 44 with ICM (30 with prior STEMI) and 20 with NICM (11 dilated, 7 hypertrophic, and 2 other types). The mean LVEF was 44.9% ± 15.2%, with 17 patients (26.6%) having an LVEF ≤35%.
Quantitative LGE analysis demonstrated that dense scar and grey zone occupied 3.6 (1.5, 7.8)% and 7.3 (2.8, 13.7)% of left ventricular myocardial tissue, respectively. Detailed characteristics of the LGE-positive patients are summarised in Table 2.
Table 2.
Baseline characteristics of the LGE-positive patients.
| Characteristic | Event in LGE-positive (n = 15) | Event-free in LGE-positive (n = 49) | P |
|---|---|---|---|
| Female | 1 (6.7%) | 9 (18.4%) | 0.275 |
| Age (years) | 56.0 (53.5,68.5) | 62.0 (55.0,69.0) | 0.461 |
| Body surface area (m2) | 1.91 ± 0.12 | 1.87 ± 0.15 | 0.374 |
| Ventricular arrhythmia Burden % | 0.3 (0.2, 0.95) | 0.3 (0.2, 2.5) | 0.904 |
| Diabetes | 4 (30.8%) | 14 (28.6%) | 0.877 |
| Hypertension | 5 (38.5%) | 24 (49.0%) | 0.499 |
| Ischemic cardiomyopathy | 9 (60.0%) | 35 (71.4%) | 0.403 |
| NYHA function class | |||
| I | 1 (6.7%) | 12 (24.5%) | 0.130 |
| II | 8 (53.3%) | 27 (55.1%) | |
| III | 5 (33.3%) | 10 (20.4%) | |
| IV | 1 (6.7%) | 0 (0%) | |
| Medications | |||
| β-blocker | 15 (100%) | 38 (77.6%) | 0.044 |
| ACEI/ARB | 15 (100%) | 37 (75.5%) | 0.033 |
| Aldosterone antagonist | 14 (93.3%) | 18 (36.7%) | <0.001 |
| Loop diuretic | 5 (33.3%) | 6 (12.2%) | 0.058 |
| LGE-CMR | |||
| LVEF, % | 35.6 ± 13.3 | 47.6 ± 14.8 | 0.008 |
| LVEF ≤35% | 7 (50.0%) | 10 (20.4%) | 0.028 |
| LVEF>50% | 1 (7.1%) | 23 (46.9%) | 0.007 |
| LV mass (g) | 153.85 (128.63, 174.1) | 143.8 (112.8, 173.75) | 0.710 |
| RVEF (%) | 50.3 ± 16.7 | 53.8 ± 9.8 | 0.408 |
| grey zone% | 14.0 (10.9, 14.7) | 5.7 (2.3, 9.2) | <0.001 |
| core scar% | 10.0 (5.9, 14.8) | 2.2 (1.1, 4.7) | <0.001 |
| Simulation | |||
| VT Inducibility | 50 (22.9, 66.9) | 0 (0, 21.1) | <0.001 |
| VT circuits | 2 (2, 4) | 0 (0, 2) | <0.001 |
| Integrated parameter | |||
| LGE-VTsim index | 69.0 (36.1, 86.5) | 6.19 (3.0, 40.56) | <0.001 |
ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; LGE, late gadolinium enhancement; LV, left ventricular; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; RVEF, right ventricular ejection fraction; VT, ventricular tachycardia. N/A: Not applicable; P values were derived from Fisher's exact test.
No significant differences in scar quantification or distribution were observed between patients with ICM (n = 45) and NICM (n = 20), likely reflecting heterogeneity within both groups. The distinction between ischaemic and non-ischaemic cardiomyopathy, together with their respective scar patterns, is illustrated in the Supplementary Materials.
The mean simulated VT induction rate, reflecting simulated electrophysiological propensity, was 5.6 (0.0, 37.5) %, with a mean of 1 (0, 2) inducible VTs per patient. To assess the clinical relevance of simulated sustained VT events, simulation outputs were compared with real-world electrophysiological and ablation findings. Among the 5 patients with VT who underwent EPS and catheter ablation, the real-world alignment with virtual VT circuits was as follows: In Patient 1, the clinical VT was non-inducible during EPS; however, the exit site identified by isochronal late activation mapping and pace mapping matched one of the virtual VT circuits. In Patients 2–4, each presented with a single monomorphic VT; the actual VT circuits correlated with the predicted virtual VT circuits, and ablation guided by these sites was successful. All 4 of these patients remained free of recurrent VT. In Patient 5, two VTs were induced during EPS, both of which aligned with 2 of the 5 virtual VT circuits identified. This patient experienced VT recurrence one week after ablation, with a morphology different from the previously documented VTs. This comparison suggests that virtual VT simulations can reflect actual arrhythmogenic substrates in patients with sustained VT and should be interpreted as probabilistic estimates of individualised risk.
3.3. Discriminatory performance of LVEF for sustained VT
In the overall population (n = 82), both LVEF ≤35% (P = 0.016) and lower LVEF as a continuous variable (P = 0.008) were significantly associated with sustained VT (Table 1). LVEF demonstrated moderate discriminatory performance, with a C-statistic of 0.730 (95% CI: 0.597–0.863). The conventional LVEF threshold of ≤35% showed limited sensitivity (50.0%) despite relatively high specificity (80.6%), whereas the Youden-derived optimal cut-off of 50.4% improved sensitivity (92.9%) at the expense of lower specificity (46.3%).
Among LGE-positive patients, those with sustained VT had significantly lower LVEF than those without sustained VT (35.6% ± 13.3% vs 45.4% ± 14.4%, P = 0.008), with a corresponding C-statistic of 0.730 (95% CI: 0.592–0.867).
3.4. Discriminatory value of visually assessed LGE for sustained VT
All patients with sustained VT were LGE-positive, whereas no VT events occurred in LGE-negative patients during the peri-CMR period, consistent with previous studies (19). Although visual LGE positivity differed significantly between patients with and without sustained VT (P = 0.023; Table 1), its discriminatory performance was modest (C-statistic: 0.634, 95% CI: 0.500–0.769) and did not differ significantly from that of LVEF (DeLong P = 0.200). These findings indicate that visual LGE assessment alone is insufficient for precise VT risk stratification and support the need for more advanced assessment among LGE-positive patients.
3.5. Development of an integrated LGE-VTsim Index
To construct an integrated discriminative model, scar and simulation parameters were compared between patients with and without sustained VT. The Event Group exhibited significantly larger core scar (10.0 (5.9, 14.8)% vs 2.2 (1.1, 4.7)%; P < 0.001 and grey zone [14.0 (10.9, 14.7)% vs 5.7 (2.3, 9.2)%; P < 0.001], higher virtual VT induction rates [50 (22.9, 66.9)% vs 0 (0, 21.1)%, P < 0.001], and a greater number of inducible VTs [2 (2, 4)vs. 0 (0, 2), P < 0.001].
Given the strong intercorrelations among these parameters (Spearman coefficients >0.7, Supplementary Table S2), EFA was performed to develop the LGE-VTsim index to reduce data dimensionality. The suitability of the data for factor analysis was confirmed by the Kaiser–Meyer–Olkin (KMO) measure (KMO = 0.742) and Bartlett's test (P < 0.001). LGE-VTsim index = 0.862 × Grey Zone + 0.614 × Core Scar + 0.906 × VT Inducibility + 0.913 × VT Circuits.
Among LGE-positive patients, the LGE-VTsim index achieved a C-statistic of 0.816 (95% CI: 0.699–0.933) for identifying sustained VT, Of which, At a cut-off value >25.6, sensitivity was 85.7% (95% CI: 57.2%–98.2%) and specificity was 71.4% (95% CI: 56.7%–83.4%) (Figure 2A). The index remained independently associated with sustained VT after adjustment for LVEF (OR: 1.029, 95% CI: 1.008–1.051; P = 0.007), whereas LVEF lost statistical significance (OR: 0.952, 95% CI: 0.903–1.003; P = 0.064). Notably, when analysed as a dichotomous variable (LGE-VTsim >25.6 vs ≤25.6), logistic regression demonstrated a markedly stronger association (regression coefficient = 2.708, OR = 15.000, 95% CI: 2.968–75.810, P < 0.001). These findings indicate the incremental discriminative value of LGE-VTsim index beyond LVEF alone.
Figure 2.
LGE–vTsim index for differentiating sustained VT events. (A) The ROC analysis of the novel LGE-VTsim index (pink, C-statistics = 0.816) and the conventional LVEF metric (teal curve, C-statistics = 0.730) in discriminating sustained VT events. Optimal cutoffs (determined by the Youden's index) are marked: 25.6 for LGE-VTsim and 50.4 for LVEF. (B) Calibration curve (with loess smoothing) of the LGE-VTsim index, illustrating the agreement between predicted probabilities of sustained VT and observed event frequencies. The dashed line represents perfect calibration (where predicted probability = observed frequency), and the shaded area denotes the 95% confidence interval reflecting estimation uncertainty. (C) Decision curve analysis comparing the net benefit of using the LGE-VTsim index (red line), LVEF (teal line), for predicting sustained VT, relative to a “None” strategy (horizontal black dashed line). The x-axis represents predicted probability thresholds, and the y-axis represents net benefit.
ROC, receiver operating characteristic; LVEF, left ventricular ejection fraction; C-index, concordance index; LGE-VTsim index = 0.862× Grey Zone + 0.614 × Core Scar + 0.906× VT Inducibility + 0.913 × VT circuits.
To assess the risk of overfitting, internal validation was performed using 1,000 bootstrap samples. The optimism-corrected calibration slope was 1.017, and the optimism-corrected C-statistic was 0.814 (original C-statistic: 0.816), indicating that the model maintained good discrimination without evidence of overfitting.
In addition, Locally Weighted Scatterplot Smoothing was applied to visualize the relationship between predicted probabilities and observed frequencies. The calibration plot (Figure 2B) demonstrated agreement between predicted and observed risks. The model achieved a Mean Absolute Error of 0.047 and a Mean Squared Error of 0.00366. An Mean Absolute Error below 0.05 indicates that, on average, predictions deviated from actual outcomes by less than 5%. The ideal calibration line remained within this interval across most of the risk spectrum, supporting the reliability of the model.
3.6. Incremental discriminative value of the LGE-VTsim Index over LVEF
Although the LGE-VTsim index demonstrated a numerically higher C-statistic compared with LVEF alone, DeLong's test (P = 0.352) did not reveal a statistically significant improvement, likely owing to the limited sample size. DCA (Figure 2C) showed a trend toward increased net benefit with LGE-VTsim index across a range of threshold probabilities.
To further evaluate the incremental value, model comparison was performed between LVEF alone and a model incorporating both LVEF and the LGE-VTsim index. Adding the LGE-VTsim index improved overall discrimination; However, not all improvement metrics reached statistical significance. Model comparisons demonstrated a ΔC of 0.093 (95% CI: −0.199–0.014, P = 0.089) and an NRI of 70.0% (95% CI: 0.116–1.284, P = 0.019). A statistically significant IDI of 15.23% was also observed (95% CI: 0.039–0.266, P = 0.008).
3.7. Differences between sustained and non-sustained VAs
The ability of the LGE-VTsim index to discriminate between sustained and non-sustained VT was further evaluated among the 36 patients with documented VAs (21 non-sustained and 15 sustained). While neither visual LGE assessment nor LVEF significantly differentiated between the two groups, quantitative LGE parameters, simulation metrics, and the LGE-VTsim index all demonstrated significant discriminatory capacity (Table 3). These findings indicate that not all VAs reflect equivalent underlying substrate severity or susceptibility to sustained episodes, thereby supporting the clinical utility of the LGE-VTsim index for refined risk stratification in this population.
Table 3.
Characteristics between sustained VT vs non-sustained VA.
| Characteristic | Sustained VT (n = 15) | Non-sustained VA (n = 21) | P |
|---|---|---|---|
| Female (%) | 1 (6.7%) | 3 (14.3%) | 0.626 |
| Age (years) | 56.0 (53.5, 68.5) | 62.0 (53.0, 69.0) | 0.835 |
| Body surface area (m2) | 1.91 ± 0.12 | 1.87 ± 0.15 | 0.398 |
| Ventricular arrhythmia Burden % | 0.3 (0.2, 0.95) | 5 (2, 17.8) | < 0.001 |
| Diabetes | 4 (30.8%) | 5 (23.8%) | 0.655 |
| Hypertension | 5 (38.5%) | 9 (42.9%) | 0.800 |
| Ischemic cardiomyopathy | 9 (60.0%) | 12 (57.1%) | 0.864 |
| NYHA function class | |||
| I | 1 (6.7%) | 4 (19.0%) | 0.858 |
| II | 8 (53.3%) | 10 (47.6%) | |
| III | 5 (33.3%) | 6 (28.6%) | |
| IV | 1 (6.7%) | 1 (4.8%) | |
| Medications | |||
| β-blocker | 15 (100%) | 18 (85.7%) | 0.250 |
| ACEI/ARB | 15 (100%) | 16 (76.2%) | 0.062 |
| Aldosterone antagonist | 14 (93.3%) | 12 (57.1%) | 0.017 |
| Loop diuretic | 5 (33.3%) | 4 (19.0%) | 0.329 |
| LGE-CMR | |||
| LVEF, % | 35.6 ± 13.3 | 43.9 ± 14.4 | 0.097 |
| LVEF≤35% (%) | 7 (46.7%) | 5 (23.8%) | 0.110 |
| LVEF>50% | 1 (6.7%) | 7 (33.3%) | 0.108 |
| LV mass (g) | 153.9 (128.6, 174.1) | 147.3 (118, 184.4) | 0.853 |
| RVEF (%) | 50.3 ± 16.7 | 51.7 ± 11.6 | 0.810 |
| LGE-positive | 15 (100%) | 19 (90.5%) | 0.500 |
| grey zone % | 14.0 (10.9, 14.7) | 7.3 (3.1, 13.8) | 0.018 |
| core scar % | 10.0 (5.9, 14.8) | 3.0 (1.4, 5.9) | <0.001 |
| Simulation | |||
| VT inducibility | 50.0 (22.9, 66.9) | 12.5 (0.0, 31.3) | 0.020 |
| inducible VT circuits | 2 (2, 4) | 1 (0, 3) | 0.039 |
| Integrated parameter | |||
| LGE-VT sim index | 69.0 (36.1, 86.5) | 20.0 (4.0, 50.3) | 0.024 |
ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; LGE, late gadolinium enhancement; LV, left ventricular; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; RVEF, right ventricular ejection fraction; VA, ventricular arrhythmia; VT, ventricular tachycardia. N/A: Not applicable; P values were derived from Fisher's exact test or independent samples t-test as appropriate.
4. Discussion
This study developed a novel LGE-VTsim index that integrates quantitative LGE-CMR scar characteristics with computational VT simulation parameters. The index demonstrated promising performance beyond LVEF for identifying sustained VT in Chinese patients with SHD. The LGE-VTsim index also effectively differentiated sustained VT from non-sustained VAs.
Current reliance on an LVEF ≤35% threshold to guide ICD therapy has well-documented limitations. Approximately 60%–70% of SCD events occur in patients with LVEF above this threshold (3, 20, 21), and many individuals with reduced LVEF may never derive benefit from ICD implantation. For example, in the conventional therapy group of the MADIT-II trial, the SCD rate was 10.0% (22, 23), which was higher than in the ICD group (3.8%), yet indicating that the vast majority of patients did not experience fatal arrhythmias. This observation highlights the limitations of a uniform LVEF-based threshold. Consistent with this, our results confirmed the suboptimal discriminatory performance of LVEF, with an optimal cut-off of ≤50.4%, suggesting that patients with higher values may be considered at low risk. The conventional LVEF ≤35% threshold was retained in LGE-negative individuals, in accordance with current guideline recommendations.
Invasive electrophysiological study (EPS) can aid risk stratification, particularly in patients with relatively preserved LVEF. The PRESERVE-EF study demonstrated that EPS effectively identified high-risk patients in a chronic ICM population with LVEF ≥40%, as malignant arrhythmias occurred exclusively in those with inducible VT (24). Although invasive EPS provides valuable prognostic information, non-invasive alternatives remain desirable.
In this context, LGE-CMR has emerged as a powerful tool for substrate-based risk assessment across various SHD aetiologies. In ICM, LGE-quantified infarct size independently predicts arrhythmic events (8), while in NICM, the presence of LGE is similarly associated with increased mortality and arrhythmic risk, as demonstrated in the DANISH-MRI sub-study (9). However, in our population, visual LGE assessment alone showed only modest discriminatory performance, and its C-statistic did not exceed that of LVEF. This finding underscores the limitations of qualitative evaluation and highlights the need for more individualized, quantitative scar analysis.
Computational VT simulation, based on patient-specific LGE-CMR–derived digital twin models, operates on principles analogous to invasive EPS but enables exhaustive pacing from multiple virtual sites to systematically assess arrhythmia susceptibility. This technique has been validated across diverse SHD populations, including HCM (25), cardiac sarcoidosis (26), tetralogy of Fallot (27), and post-infarction patients (27), with simulated VT inducibility consistently correlating with clinical outcomes. The present study extends these observations by integrating quantitative LGE characteristics (core scar, grey zone) with simulation-derived parameters (VT inducibility rate, number of circuits) into a unified index. The LGE-VTsim index demonstrated significantly improved discrimination for sustained VT compared with LVEF alone and remained independently associated with sustained VT after adjustment for LVEF.
Importantly, the LGE-VTsim index also demonstrated the ability to differentiate non-sustained VAs from sustained VT. Although the Improve SCA study identified frequent PVCs as a risk factor for SCD in primary prevention (28), our findings revealed significant differences in both LGE characteristics and in silico VT inducibility between these patient groups. Notably, the vast majority (13 out of 14) of patients with sustained VT showed no evidence of frequent PVCs on electrocardiographic monitoring. These results suggest that not all VAs confer equivalent clinical risk and highlight the potential of the LGE-VTsim index to improve risk stratification by identifying patients with a truly high-risk arrhythmogenic substrate.
Based on the present findings, the LGE VTsim index represents a promising hypothesis-generating tool for sustained VT risk identification in patients with suspected SHD (Figure 3). Initial risk categorisation incorporated LVEF and LGE-VTsim index status: the combination of LGE-negative status and LVEF >50.4% identified patients at low risk, whereas the conventional LVEF ≤35% threshold served as a high-risk marker owing to its high specificity. Among LGE-positive patients, for whom visual assessment showed limited accuracy, the LGE-VTsim index effectively identified high-risk individuals and distinguished sustained from non-sustained VAs. At a cut-off value of >25.6, the index demonstrated high sensitivity and good specificity. What's more, advances in machine learning models leveraging ECG, imaging, and clinical data showed promises to improve arrhythmic risk prediction (29). However, corresponding regulatory frameworks are yet to be established for AI guided arrhythmic risk prediction.
Figure 3.
Sustained VT risk discrimination based on LGE-VTsim index. This flowchart presents an exploratory stepwise framework for VT risk stratification in suspected SHD. LGE-CMR screening first divides patients: LGE-positive cases undergo digital twin reconstruction and VT simulation to derive an LGE-VTsim index (>25.6 indicating high risk), while LGE-negative cases are stratified by LVEF (≤35 indicating high risk as conventionally suggested). This integrated approach enhances risk recognition beyond LVEF alone.
CMR, cardiac magnetic resonance; LGE, late gadolinium enhancement; LVEF, left ventricular ejection fraction; SHD, structural heart disease; VT, ventricular tachycardia.
To our knowledge, this is the first study to explore the feasibility of combining LGE-CMR with computational simulation in a Chinese population. The exploratory framework employs LGE-CMR for initial screening and applies the LGE-VTsim index for advanced assessment in LGE-positive patients, providing a pathophysiology-informed and individualised approach to VT assessment (Figure 4). This parameter appears particularly valuable for optimising risk evaluation in patients with LVEF between 35% and 50.4%, a population for whom current clinical guidelines offer limited specific guidance.
Figure 4.
Central Illustration. This study developed a hypothesis-generating parameter, named LGE-VTsim index, incorporating LGE-CMR for scar quantification, and virtual VT inducibility and circuits. the proposed method overcomes limitations of traditional LVEF-based risk assessment.
LGE-CMR, late gadolinium enhancement cardiac magnetic resonance; VT, ventricular tachycardia; SCD, sudden cardiac death; SHD, structural heart disease; LVEF, left ventricular ejection fraction.
5. Limitations
Several limitations should be acknowledged. First, the single-centre design and limited sample size increase the risk of overfitting, necessitating external validation in larger, multicentre aetiology-stratified cohort. Future studies should incorporate inter-observer reproducibility and compare findings with invasive electrophysiological study to confirm physiological relevance. Second, the 6-month arrhythmia window and a single baseline LGE-CMR examination may capture disease progression; serial CMR with updated LGE-VTsim index calculations longitudinal risk evaluation. Third, the cohort was predominantly male (83.1%), potentially limiting generalisability to women. Sex-based differences in myocardial scarring—with women exhibiting less extensive fibrosis—may contribute to lower malignant arrhythmia rates and underscore the need for sex-sensitive risk stratification beyond LVEF (30, 31), pending future HCM specific validation (32), Finally, although β-blocker therapy was not adjusted for in the model, its widespread use among patients with sustained VT supports the model's applicability to treated populations.
6. Conclusion
This study demonstrates the feasibility of combining LGE-CMR with computational VT simulation in a Chinese population and proposes the LGE-VTsim index as a hypothesis-generating parameter for identifying sustained VT in SHD. Compared with LVEF alone, the LGE-VTsim index showed improved trend of discriminative performance over LVEF alone and effectively distinguished sustained from non-sustained arrhythmias.
Acknowledgements
The authors thank the patients for their valued contributions to this study.
Funding Statement
The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the National Natural Science Foundation of China (82100334 to Kun Zuo) and the Beijing Municipal Administration of Hospitals’ Youth Programme (QML20170303 to Kun Zuo), the Fundamental Research Funds for the Central Universities (DUT25YG231 to Dongdong Deng).
Footnotes
Edited by: Aikeliyaer Ainiwaer, Maastricht University, Netherlands
Reviewed by: Nicola Pierucci, Sapienza University of Rome, Italy
Damiano Fedele, University of Bologna, Italy
Abbreviations ARVC, arrhythmogenic right ventricular cardiomyopathy; CMR, cardiac magnetic resonance; DCM, dilated cardiomyopathy; EPS, electrophysiological study; HCM, hypertrophic cardiomyopathy; ICD, implantable cardioverter-defibrillator; ICM, ischemic cardiomyopathy; LGE, late gadolinium enhancement; LVEF, left ventricular ejection fraction; NICM, non-ischemic cardiomyopathy; PVC, premature ventricular contraction; SCD, sudden cardiac death; SHD, structural heart disease; STEMI, ST-segment elevation myocardial infarction; NSTEMI, non-ST-segment elevation myocardial infarction; VA, ventricular arrhythmia; VT, ventricular tachycardia.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethics statement
The studies involving humans were approved by Ethics Committee of Beijing Chaoyang Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
Author contributions
KZ: Conceptualization, Funding acquisition, Investigation, Project administration, Writing – original draft, Writing – review & editing. KL: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. LX: Methodology, Software, Writing – original draft, Writing – review & editing. YW: Methodology, Software, Writing – original draft, Writing – review & editing. SL: Investigation, Methodology, Software, Writing – original draft, Writing – review & editing. ZK: Methodology, Software, Writing – original draft, Writing – review & editing. ZF: Methodology, Software, Writing – original draft, Writing – review & editing. JZ: Conceptualization, Writing – original draft, Writing – review & editing. RD: Software, Writing – original draft, Writing – review & editing. LX: Data curation, Methodology, Resources, Software, Writing – original draft, Writing – review & editing. DD: Data curation, Investigation, Methodology, Resources, Software, Writing – original draft, Writing – review & editing. ZL: Conceptualization, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcvm.2026.1870223/full#supplementary-material
References
- 1.Leyva F, Israel CW, Singh J. Declining risk of sudden cardiac death in heart failure: fact or myth? Circulation. (2023) 147(9):759–67. 10.1161/CIRCULATIONAHA.122.062159 [DOI] [PubMed] [Google Scholar]
- 2.Moss AJ, Greenberg H, Case RB, Zareba W, Hall WJ, Brown MW, et al. Long-term clinical course of patients after termination of ventricular tachyarrhythmia by an implanted defibrillator. Circulation. (2004) 110(25):3760–5. 10.1161/01.CIR.0000150390.04704.B7 [DOI] [PubMed] [Google Scholar]
- 3.Mäkikallio TH, Barthel P, Schneider R, Bauer A, Tapanainen JM, Tulppo MP, et al. Prediction of sudden cardiac death after acute myocardial infarction: role of holter monitoring in the modern treatment era. Eur Heart J. (2005) 26(8):762–9. 10.1093/eurheartj/ehi188 [DOI] [PubMed] [Google Scholar]
- 4.Al-Khatib SM, Stevenson WG, Ackerman MJ, Bryant WJ, Callans DJ, Curtis AB, et al. 2017 AHA/ACC/HRS guideline for management of patients with ventricular arrhythmias and the prevention of sudden cardiac death: executive summary: a report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines and the heart rhythm society. Circulation. (2018) 138(13):e210–71. 10.1161/CIR.0000000000000548 [DOI] [PubMed] [Google Scholar]
- 5.de Bakker JM, van Capelle FJ, Janse MJ, Wilde AA, Coronel R, Becker AE, et al. Reentry as a cause of ventricular tachycardia in patients with chronic ischemic heart disease: electrophysiologic and anatomic correlation. Circulation. (1988) 77(3):589–606. 10.1161/01.CIR.77.3.589 [DOI] [PubMed] [Google Scholar]
- 6.Harada T, Stevenson WG, Kocovic DZ, Friedman PL. Catheter ablation of ventricular tachycardia after myocardial infarction: relation of endocardial sinus rhythm late potentials to the reentry circuit. J Am Coll Cardiol. (1997) 30(4):1015–23. 10.1016/S0735-1097(97)00257-X [DOI] [PubMed] [Google Scholar]
- 7.Mountantonakis SE, Park RE, Frankel DS, Hutchinson MD, Dixit S, Cooper J, et al. Relationship between voltage map “channels” and the location of critical isthmus sites in patients with post-infarction cardiomyopathy and ventricular tachycardia. J Am Coll Cardiol. (2013) 61(20):2088–95. 10.1016/j.jacc.2013.02.031 [DOI] [PubMed] [Google Scholar]
- 8.Arbustini E, Kramer CM, Narula J. Arrhythmogenic potential of border zone after myocardial infarction: scar is more than just a healed wound. JACC Cardiovasc Imaging. (2018) 11(4):573–6. 10.1016/j.jcmg.2017.07.003 [DOI] [PubMed] [Google Scholar]
- 9.Elming MB, Hammer-Hansen S, Voges I, Nyktari E, Raja AA, Svendsen JH, et al. Myocardial fibrosis and the effect of primary prophylactic defibrillator implantation in patients with non-ischemic systolic heart failure-DANISH-MRI. Am Heart J. (2020) 221:165–76. 10.1016/j.ahj.2019.10.020 [DOI] [PubMed] [Google Scholar]
- 10.Oshinski JN, Yang Z, Jones JR, Mata JF, French BA. Imaging time after gd-DTPA injection is critical in using delayed enhancement to determine infarct size accurately with magnetic resonance imaging. Circulation. (2001) 104(23):2838–42. 10.1161/hc4801.100351 [DOI] [PubMed] [Google Scholar]
- 11.Amado LC, Gerber BL, Gupta SN, Rettmann DW, Szarf G, Schock R, et al. Accurate and objective infarct sizing by contrast-enhanced magnetic resonance imaging in a canine myocardial infarction model. J Am Coll Cardiol. (2004) 44(12):2383–9. 10.1016/j.jacc.2004.09.020 [DOI] [PubMed] [Google Scholar]
- 12.Schmidt A, Azevedo CF, Cheng A, Gupta SN, Bluemke DA, Foo TK, et al. Infarct tissue heterogeneity by magnetic resonance imaging identifies enhanced cardiac arrhythmia susceptibility in patients with left ventricular dysfunction. Circulation. (2007) 115(15):2006–14. 10.1161/CIRCULATIONAHA.106.653568 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hammersley DJ, Zegard A, Androulakis E, Jones RE, Okafor O, Hatipoglu S, et al. Arrhythmic risk stratification by cardiovascular magnetic resonance imaging in patients with nonischemic cardiomyopathy. J Am Coll Cardiol. (2024) 84(15):1407–20. 10.1016/j.jacc.2024.06.046 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Del Franco A, Ruggieri R, Pieroni M, Ciabatti M, Zocchi C, Biagioni G, et al. Atlas of regional left ventricular scar in nonischemic cardiomyopathies: substrates and etiologies. JACC Adv. (2024) 3(10):101214. 10.1016/j.jacadv.2024.101214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Aziz Z, Shatz D, Raiman M, Upadhyay GA, Beaser AD, Besser SA, et al. Targeted ablation of ventricular tachycardia guided by wavefront discontinuities during Sinus rhythm: a new functional substrate mapping strategy. Circulation. (2019) 140(17):1383–97. 10.1161/CIRCULATIONAHA.119.042423 [DOI] [PubMed] [Google Scholar]
- 16.Andreini D, Dello Russo A, Pontone G, Mushtaq S, Conte E, Perchinunno M, et al. CMR For identifying the substrate of ventricular arrhythmia in patients with normal echocardiography. JACC Cardiovasc Imaging. (2020) 13(2 Pt 1):410–21. 10.1016/j.jcmg.2019.04.023 [DOI] [PubMed] [Google Scholar]
- 17.Williams CB, Andrade JG, Hawkins NM, Cheung C, Krahn A, Laksman ZW, et al. Establishing reference ranges for ambulatory electrocardiography parameters: meta-analysis. Heart. (2020) 106(22):1732–9. 10.1136/heartjnl-2020-316925 [DOI] [PubMed] [Google Scholar]
- 18.van der Ree MH, Visser J, Planken RN, Dieleman EMT, Boekholdt SM, Balgobind BV, et al. Standardizing the cardiac radioablation targeting workflow: enabling semi-automated angulation and segmentation of the heart according to the American Heart Association segmented model. Adv Radiat Oncol. (2022) 7(4):100928. 10.1016/j.adro.2022.100928 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Prasad SK, Akbari T, Bishop MJ, Halliday BP, Leyva-Leon F, Marchlinski F. Late gadolinium enhancement imaging and sudden cardiac death. Eur Heart J. (2025) 46(36):3555–68. 10.1093/eurheartj/ehaf464 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gorgels AP, Gijsbers C, de Vreede-Swagemakers J, Lousberg A, Wellens HJ. Out-of-hospital cardiac arrest–the relevance of heart failure. The Maastricht circulatory arrest registry. Eur Heart J. (2003) 24(13):1204–9. 10.1016/S0195-668X(03)00191-X [DOI] [PubMed] [Google Scholar]
- 21.Dagres N, Hindricks G. Risk stratification after myocardial infarction: is left ventricular ejection fraction enough to prevent sudden cardiac death? Eur Heart J. (2013) 34(26):1964–71. 10.1093/eurheartj/eht109 [DOI] [PubMed] [Google Scholar]
- 22.Greenberg H, Case RB, Moss AJ, Brown MW, Carroll ER, Andrews ML. Analysis of mortality events in the multicenter automatic defibrillator implantation trial (MADIT-II). J Am Coll Cardiol. (2004) 43(8):1459–65. 10.1016/j.jacc.2003.11.038 [DOI] [PubMed] [Google Scholar]
- 23.Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS, et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med. (2002) 346(12):877–83. 10.1056/NEJMoa013474 [DOI] [PubMed] [Google Scholar]
- 24.Gatzoulis KA, Tsiachris D, Arsenos P, Antoniou C-K, Dilaveris P, Sideris S, et al. Arrhythmic risk stratification in post-myocardial infarction patients with preserved ejection fraction: the PRESERVE EF study. Eur Heart J. (2019) 40(35):2940–9. 10.1093/eurheartj/ehz260 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Shade JK, Prakosa A, Popescu DM, Yu R, Okada DR, Chrispin J, et al. Predicting risk of sudden cardiac death in patients with cardiac sarcoidosis using multimodality imaging and personalized heart modeling in a multivariable classifier. Sci Adv. (2021) 7(31):eabi8020. 10.1126/sciadv.abi8020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shade JK, Cartoski MJ, Nikolov P, Prakosa A, Doshi A, Binka E, et al. Ventricular arrhythmia risk prediction in repaired tetralogy of fallot using personalized computational cardiac models. Heart Rhythm. (2020) 17(3):408–14. 10.1016/j.hrthm.2019.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Arevalo HJ, Vadakkumpadan F, Guallar E, Jebb A, Malamas P, Wu KC, et al. Arrhythmia risk stratification of patients after myocardial infarction using personalized heart models. Nat Commun. (2016) 7:11437. 10.1038/ncomms11437 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhang S, Ching CK, Huang D, Liu Y-B, Rodriguez-Guerrero DA, Hussin A, et al. Utilization of implantable cardioverter-defibrillators for the prevention of sudden cardiac death in emerging countries: improve SCA clinical trial. Heart Rhythm. (2020) 17(3):468–75. 10.1016/j.hrthm.2019.09.023 [DOI] [PubMed] [Google Scholar]
- 29.Cipollone P, Pierucci N, Matteucci A, Palombi M, Laviola D, Bruti R, et al. Artificial intelligence in cardiac electrophysiology: a comprehensive review. J Pers Med. (2025) 15(11):532. 10.3390/jpm15110532 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Regitz-Zagrosek V, Gebhard C. Gender medicine: effects of sex and gender on cardiovascular disease manifestation and outcomes. Nat Rev Cardiol. (2023) 20(4):236–47. 10.1038/s41569-022-00797-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Narasimha D, Curtis AB. Sex differences in utilisation and response to implantable device therapy. Arrhythm Electrophysiol Rev. (2015) 4(2):129–35. 10.15420/AER.2015.04.02.129 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Guglielmo M, Fedele D, Bergamaschi L, Armillotta M, Angeli F, Ciarlantini M, et al. Cardiac magnetic resonance-derived left atrioventricular coupling index predicts outcome in reduced ejection fraction. ESC Heart Fail. (2026) 13(3):xvag130. 10.1093/eschf/xvag130 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.