European Heart Journal - Digital Health

Large epidemiologic studies have historical electrocardiogram (ECG) collections limited to paper tracings or scanned images rather than native digital files, limiting ECG-AI applications. We compared 15 ECGs available as native digital XML files and as PDF tracings digitized using ECGScan software. Both versions were processed using an ensemble of five convolutional neural networks predicting 10-year heart failure risk. Predictions were strongly correlated (Pearson r=0.804; Spearman {rho}=0.893). Predicted probabilities were [mean (SD)] 0.195 (0.048) for native digital ECGs and 0.171 (0.042) for digitized ECGs. These preliminary findings support further evaluation of digitized ECGs for ECG-AI applications.

Background: Rheumatic mitral stenosis (RMS) disproportionately affects low-resource settings where echocardiography is limited. We evaluated foundation model (FM) feasibility for RMS detection using limited, indirectly labeled ECGs. Methods: An ECG-FM was fine-tuned using 804 RMS-positive ECGs per fold and externally validated across clinical and wearable lead configurations. Results: Internal AUROCs were 0.89, 0.87, and 0.87, for 12-Lead, 6-Lead, and 1-Lead respectively. External AUROCs were 0.82, 0.80, and 0.77. Performance remained comparable across demographic subgroups. Conclusions: FM fine-tuning enables RMS detection using limited ICD-derived labels, supporting further evaluation of reduced-lead ECG screening and echocardiography triage in resource-constrained settings.

Background: Artificial intelligence (AI) may improve electrocardiographic differentiation of Takotsubo syndrome from acute anterior ST-segment elevation myocardial infarction (STEMI), but its impact on human diagnostic performance, the added value of visual explanations, and differences by reader background and experience remain uncertain. Methods: Using 12-lead electrocardiogram (ECG) images from two hospitals, we developed and externally validated an ensemble of three deep learning models after self-supervised pretraining and fine-tuning. The diagnostic task used ECGs recorded within 14 days after symptom onset, regardless of reperfusion status. We conducted a four-phase reader study. Phases 1-3 compared unaided interpretation, a reference sheet, and additional AI assistance. In Phase 4, readers evaluated a new ECG set after randomization to reference-sheet support alone, with AI predictions, or with AI predictions plus gradient-weighted class activation mapping (Grad-CAM) heatmaps. Diagnostic correctness was analyzed using generalized linear mixed-effects models. Exploratory analyses examined differences by cardiologist status and experience. Results: Development and external-validation cohorts, including healthy controls, comprised 832 and 95 participants, respectively. For Takotsubo-STEMI discrimination, external-validation accuracy was 0.914, with areas under the receiver operating characteristic and precision-recall curves of 0.970 and 0.984. Seventy healthcare providers completed the reader study. The estimated probability of correct diagnosis increased from 0.632 without assistance to 0.864 with AI assistance (odds ratio [OR], 3.71; P<0.001). In randomized Phase 4, AI assistance improved accuracy vs reference-sheet support alone (OR, 4.28; 95% CI, 3.08-5.95; P<0.001), and Grad-CAM provided additional benefit (OR, 1.53; 95% CI, 1.04-2.25; P=0.033). Exploratory analyses showed greater improvement among less experienced readers; with AI assistance, non-cardiologists outperformed cardiologists. AI-assisted readers remained less accurate overall than the model alone. Conclusions: An ECG image-based ensemble differentiated Takotsubo syndrome from anterior STEMI and improved reader accuracy. Visual explanations provided additional benefit, while experience-related differences highlight the importance of effective human-AI interaction.

Background. Real-world 12-lead electrocardiograms (ECGs) often have missing or corrupted leads. A deep-learning classifier can be prepared for this by random lead masking during training, or by injecting clinical knowledge, for example by pulling its attention towards the leads that guidelines use for each diagnosis. Whether this knowledge adds robustness beyond masking is unknown. Methods. Pre-registered comparison of three arms sharing one network (per-lead 1D-ResNet encoder and per-diagnosis attention over the 12 leads): standard training (M0); random whole-lead masking (M1); masking plus a Kullback-Leibler penalty pulling attention towards a fixed guideline-derived key-lead map (M2, guideline-based attention guidance). Six diagnoses; five seeds per arm. Development on PTB-XL (21,799 ECGs); single sealed evaluation on PTB-XL fold 10 (2,198 ECGs) and on CODE-15% (233,204 exams, external validation). Primary endpoint: AUROC difference M2 - M1, averaged over first-degree AV block, right and left bundle branch block and atrial fibrillation and over 1 to 6 lost (zeroed) leads, with a patient-level paired BCa bootstrap 95% confidence interval. Results. The guidance did not change robustness: {Delta}AUROC -0.0010 (95% CI -0.0033 to +0.0006), an interval entirely within the pre-specified +-0.02 margin; the result was unchanged with another bootstrap seed, other interval methods and leave-one-diagnosis-out analyses. Masking improved AUROC: M1 - M0 = +0.0134 (95% CI +0.0102 to +0.0172) when one key lead was lost, and the number of lost leads tolerated before AUROC fell by more than 0.02 rose from 7 (M0) to 10 (M1). M2 attention matched the map and M2 was non-inferior on intact ECGs, but its calibration slope was lower for all six diagnoses in both datasets. M2 lost slightly more when all key leads were lost together (-0.0067) and gained only when one to three leads remained (exploratory; AUROC 0.9408 vs 0.9210 with 11 leads lost). CODE-15% reproduced every pattern (primary -0.0007, 95% CI -0.0011 to -0.0003). Conclusions. Random lead masking suffices to make an ECG classifier robust to missing leads; adding guideline knowledge aligns attention with clinical criteria but adds no robustness and worsens calibration. Guideline-aligned attention should not be taken as evidence of robustness.

Aims: Myocardial strain enables sensitive assessment of cardiac function, but its clinical application has largely focused on individual chambers or selected patient populations. We evaluated the feasibility and clinical relevance of fully automated multi-chamber strain assessment in consecutive patients undergoing routine transthoracic echocardiography (TTE). Methods and results: This retrospective study included 8,350 consecutive adults undergoing TTE. A hospital-wide artificial intelligence (AI)-based system automatically measured left ventricular (LV) global longitudinal strain (LVGLS), left atrial (LA) reservoir strain (LASr), and right ventricular (RV) free-wall strain (RVFWS). AI-derived LVGLS was available from at least one apical view in 97.1% of patients and from all three views in 94.8%; LASr and RVFWS were available in 94.1% and 94.5%, respectively, resulting in complete multi-chamber strain assessment in 90.3%. Strain measures showed modest correlations with conventional chamber-specific echocardiographic parameters and differential associations with clinical outcomes. Gaussian mixture modeling identified three distinct clusters characterized by preserved multi-chamber function, RV-dominant dysfunction, and pan-chamber dysfunction. Compared with the preserved cluster, pan-chamber dysfunction was associated with the highest risk of all-cause death or heart failure hospitalization (adjusted hazard ratio, 3.44; 95% CI, 2.43-4.88; P<0.001). Prognostic associations remained evident in patients with LV ejection fraction (LVEF) [&ge;]50%, and cluster assignment provided incremental prognostic information beyond clinical factors, LVEF, and RV fractional area change. Conclusions: Fully automated multi-chamber strain assessment was feasible in a large consecutive TTE cohort. Integrating strain across cardiac chambers may provide complementary functional characterization and identify clinically relevant clusters with prognostic information beyond conventional assessment.

Background: Heart failure with preserved ejection fraction (HFpEF) accounts for approximately half of the more than 64 million heart failure cases globally, yet its pathophysiology is incompletely understood, its molecular determinants are poorly defined, and disease-specific therapies remain limited. Comprehensive characterization of the cardiac motion abnormalities central to HFpEF has not been feasible at the population level. Methods: We developed a deep learning framework integrating image segmentation with optical flow motion analysis and applied it to standard cine cardiac magnetic resonance images from 83,569 UK Biobank participants, deriving 32 myocardial and inner cavity velocity phenotypes spanning the cardiac cycle, including mid-diastolic velocities not previously quantified at population scale. We examined the prognostic relevance, genetic architecture, and candidate causal mediators of the optical flow velocity phenotypes. Results: In a pragmatically defined HFpEF subcohort, mid-diastolic and late-diastolic optical flow velocities were reduced relative to healthy reference participants, while higher systolic left ventricular myocardial velocity was associated with lower all-cause mortality (HR 0.61 per SD; 95% CI 0.45-0.82). Genome-wide association analyses identified 12 risk loci associated with optical flow velocity phenotypes. Phospholamban (PLN), the canonical regulator of sarcoplasmic reticulum calcium reuptake, demonstrated the broadest pleiotropy across all cardiac phases. SOX5, a transcription factor involved in extracellular matrix development, showed significant genome-wide association exclusively with mid-diastolic velocities. Mendelian randomization implicated RABGAP1L as a candidate causal mediator of early diastolic velocity (IVW {beta} = -0.255 per NPX for early diastolic right ventricular inner circumferential velocity; p = 5.49 x 10^-22), nominating a calcium-handling pathway in diastolic dysfunction. Conclusion: Genetic and causal-inference evidence implicates intracellular calcium handling and extracellular matrix remodeling as candidate mechanisms underlying diastolic dysfunction. Together, this work establishes deep-learning-enabled cardiac motion phenotyping as a scalable approach to interrogate the molecular basis of HFpEF and nominates candidate targets for therapeutic development.

Standard cardiac biomarkers such as ejection fraction, ventricular volumes, and strain compress complex three-dimensional heart dynamics into a small set of predefined measurements, leaving clinically important heterogeneity unrecognised. Here we propose a personalised normative cardiac framework in which each individual is compared to a digital twin representing expected healthy cardiac function given their age, sex, and body habitus. We show that structured patterns of deviation from these normative twins, discovered through systematic exploration of a generative cardiac dynamics space, constitute novel dynamic phenotypes that are not captured by conventional biomarkers. Applied independently to two UK Biobank cardiac cohorts with distinct shape representation, a 128-dimensional four-chamber statistical shape descriptor (HeartSSM, n = 52,365) and a 64-dimensional deep generative latent space from a conditional variational mesh autoencoder (MeshHeart cVAE, n = up to 39,376), we identify reproducible, continuous deviation-from-normative-twin scores (three discrete deviation phenotypes in the HeartSSM cohort) that improve cardiovascular risk stratification beyond ejection fraction, ventricular volumes, myocardial mass, and strain. The strongest effect was seen for incident atrial fibrillation among individuals with preserved ejection fraction (at least 50%): incremental C-index delta C = +0.068 (5-fold cross-validated delta C = +0.063; likelihood-ratio P = 3.3 x 10^-8), with a separate, smaller effect for incident stroke in the same preserved-EF subgroup in the independent MeshHeart cVAE representation (which showed no signal for atrial fibrillation) (delta C = +0.024, P = 2.6 x 10^-4). Both effects were checked against placebo controls in which the deviation scores were replaced by Gaussian noise. To test whether the normative twin generalises beyond UK Biobank's specific scanning and processing pipeline, we applied the MeshHeart cVAE normative twin, unmodified, to an independent external cohort of healthy volunteers imaged at a different hospital (GenScan/Hammersmith, n = 315): deviation scores in this cohort were of comparable magnitude to the UK Biobank healthy reference (ratio of means = 1.09; Mann-Whitney P = 1.4 x 10^-5), indicating that the normative twin captures cross-site-transferable structure rather than a scanner- or pipeline-specific artefact. Our findings suggest that cardiac phenotyping can move beyond predefined measurements toward data-driven dynamic phenotypes that better capture the heterogeneity underlying cardiovascular disease risk, even among individuals whose ejection fraction appears normal.

Background Patients with normal perfusion, normal myocardial flow reserve (MFR) and no coronary artery calcium (CAC) on myocardial perfusion imaging are considered low risk. Yet, adverse events still occur and no routinely reported imaging marker stratifies residual risk. This study aims to determine whether fully automated ascending aortic calcification (AAC) from the attenuation correction CT stratifies risk in these patients. Methods In the multicenter REFINE-PET registry, AAC was quantified from non-contrast CT using deep learning aortic segmentation and Agatston scoring. Triple-normal was defined as stress total perfusion deficit (TPD) <5%, MFR >2 and CAC 0. The primary endpoint was cardiovascular death, myocardial infarction or heart failure admission, with heart failure admission as key secondary endpoint. Cox models were adjusted for clinical confounders. Results Of 21,923 patients, 3,580 (16.3%) were triple-normal, of whom 1,322 (36.9%) had AAC (median Agatston 65, IQR 22-212). Agreement with three blinded readers was good to excellent (Intraclass correlation coefficient 0.87-0.90). Over a median follow-up of 52 months (IQR 26-69), 194 primary events occurred. Per standard deviation, AAC was independently associated with the primary endpoint (adjusted HR 1.24, 95% CI 1.09-1.40; p<0.001), heart failure admission (adjusted HR 1.30, 95% CI 1.10-1.55; p=0.003) and all-cause mortality (adjusted HR 1.17, 95% CI 1.04-1.32; p=0.009). AAC improved model fit beyond clinical factors, MFR and TPD (p< or =0.009), with modest gains in discrimination ({Delta}C 0.007-0.009). Conclusions In patients with normal PET readouts, automated AAC was independently associated with adverse outcomes, most strongly heart failure admission, beyond clinical risk factors, perfusion and flow.

Explicit timing and context features may complement short electrocardiogram waveforms, but aggregate gains on imbalanced benchmarks can conceal errors in individual classes. We therefore compared Logistic Regression, Random Forest, XGBoost, and a one-dimensional convolutional neural network (CNN) in four cumulative input stages. Stage S1 used a 255-sample beat waveform alone, and stages S2 to S4 added 6, 13, and 21 engineered features. Using the inter-patient division of the MIT-BIH Arrhythmia Database, with 50,992 development and 49,686 evaluation beats in four classes, we estimated 95% intervals with a paired bootstrap over the 22 evaluation recordings. Adding six RR interval features increased macro F1 by 0.025 to 0.093, with intervals excluding zero for every model family. Further context helped Random Forest and XGBoost, whereas later CNN gains fell within resampling uncertainty. At the final stage, XGBoost and CNN reached macro F1 of 0.580 and 0.574, and the interval for their difference included zero. Despite opposite operating points, their precision for fusion beats was similar at matched recall, suggesting that neither separated fusion beats better. The most accurate pipeline, Random Forest at S2 with 94.45% accuracy, detected only 1.4% of supraventricular beats, and development and evaluation favored the same stage for only one family. The results for the minority classes were based on single recordings. Record 213 supplied 203 of the 207 correct CNN fusion detections, and elsewhere both final models detected at most 4 of 26 fusion beats. Record 232 contains 75% of the evaluation supraventricular beats, and nearly all were missed. Excluding this record nearly doubled XGBoost supraventricular F1, from 0.199 to 0.383. These findings are limited to retrospective analysis with annotated beat positions and a single training seed.

Background: Patients with heart failure (HF) experience high mortality, yet survival estimates remain challenging. Available prognostic tools commonly predict survival beyond 1-year, limiting goals of care discussions and enrollment on palliative care programs. The objective of this project was to develop a novel prognostic score to better predict 6-month mortality. Methods: Patients with HF admitted to the St. Boniface Hospital (Winnipeg, Canada) between January 1, 2019 and January 1, 2024 were identified. Patient characteristics, imaging parameters, laboratory values, and functional assessments (Palliative Performance Scale; PPSv2) were collected. Patients were randomly divided into development (80%) and validation (20%) cohorts. Penalized LASSO regression was used to select variables most predictive of 6-month mortality in the development cohort. Retained variables were incorporated into a multivariable logistic regression model to derive the CARDIO-PAL score, the performance of which was assessed in the holdout validation cohort. Results: Characteristics were similar between the development (n = 2036) and validation (n = 508) cohorts, with 6-month mortality rates of 29% and 31%, respectively. 8 candidate variables were retained: age (adjusted odds ratio [aOR] 1.01), PPSv2 (aOR 2.08), systolic blood pressure (aOR 1.14), discontinued use of angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker (aOR 2.58), beta-blocker (aOR 1.59), and mineralocorticoid receptor antagonist (aOR 2.67), right ventricular systolic pressure (aOR 2.06), and urea (aOR 2.13). The CARDIO-PAL score demonstrated good discrimination, with an area under the curve of 0.80 (95% confidence interval 0.75 - 0.84) and good calibration (slope 0.84, intercept -0.04). 6-month mortality across low, intermediate, and high-risk groups was 2%, 15%, and 54%, respectively. Conclusion: The CARDIO-PAL score is a novel tool to predict 6-month mortality in HF inpatients. A collaborative approach integrating CARDIO-PAL with other prognostic scores and serial clinical assessments has the potential to promote patient-centered care and improve referral to palliative care programs.

Background This study presents reference ranges for cardiovascular magnetic resonance (CMR)-derived myocardial strain from the Healthy Hearts Consortium, while evaluating variations by demographic characteristics (age, sex, and ethnicity) and technical factors (scanner vendor, magnetic field strength). We derived reference ranges and centile-based categories for ventricular, atrial, and rapid strain parameters. The prognostic relevance of these categories was assessed by examining their associations with incident cardiovascular events. Methods The dataset included 9,088 CMR studies. Automated image analysis pipelines were established using suiteHEART (NeoSoft Inc.) and cvi42 (Circle Cardiovascular Imaging). A three-step quality control framework was applied including exclusion of incomplete data, visual assessment of image and segmentation quality, and statistical outlier removal. Ventricular strain and strain rate were measured using both software solutions. Additional cvi42 derived metrics included velocity and displacement, annular plane systolic excursion metrics, and rapid strain parameters. Age- and sex-specific reference ranges were generated using lambda-mu-sigma (LMS) modeling with penalized splines. Reference ranges were defined as values between the 2.5th and 97.5th percentiles. For lower-tail classification, values below the 2.5th percentile were considered mildly out-of-range and values below the 0.5th percentile at least moderately out-of-range. Differences across scanner vendors (Siemens, Philips, GE) and magnetic field strengths (1.5 T vs. 3 T) were evaluated using adjusted linear models. The prognostic relevance of selected cvi42-derived categories was evaluated in approximately 45,000 UK Biobank participants using Cox regression adjusted for age, sex, and body mass index. Results After quality control, 8,975 CMR examinations (cvi42) and 8,992 (suiteHEART) were available for analysis; these represent software-specific analytic cohorts. Most exclusions occurred during visual quality control. Women had higher absolute ventricular and atrial strain values than men across all age groups. Left ventricular (LV) longitudinal strain and strain rate were slightly lower in older ages, whereas circumferential and radial parameters remained largely stable. Left atrial strain was also lower in older age groups, while right atrial strain did not vary by age. Annular plane systolic excursion and rapid strain metrics showed similar sex-specific patterns, with modestly lower atrial and displacement values in older groups. Ethnicity-related differences were modest, heterogeneous, and substantially smaller in magnitude than age- and sex-related differences. Technical effects were parameter specific. While LV longitudinal and circumferential strains were largely unaffected, radial strain was higher at 3 T and showed vendor-related differences. LMS-derived reference ranges and centile thresholds were applied both across all ethnicities and within ethnicity-specific analyses, and are presented stratified by age and sex. In the prognostic analysis, centile categories were most consistently associated with incident heart failure. LV longitudinal, circumferential, and radial strain below the 0.5th centile were associated with significantly increased risk of incident heart failure (HRs: 9.9 to 12.4, 95% CIs 7.00-13.90 to 8.49-18.06). Conclusion This study provides age- and sex-specific reference ranges and centile-based categories for CMR-derived strain parameters in a large reference population. Age and sex represent the primary determinants of myocardial deformation, whereas scanner-related effects are parameter-specific. Values in the lower extremes of the healthy reference distribution, particularly for LV strain, are associated with increased risk of incident heart failure. These findings enable standardized, centile-based classification of myocardial strain, facilitating consistent interpretation and improving risk stratification in research and clinical practice.

BackgroundObstructive sleep apnea (OSA) is a highly prevalent sleep disorder. Major risk factors include obesity, older age, and male sex. Given the well-established association between OSA and cardiovascular disease, markers of cardiovascular health derived from electrocardiography (ECG) may help identify individuals with undiagnosed OSA. This study evaluated whether 12-lead ECG features improve machine learning-based prediction of OSA beyond established clinical factors. MethodsWe conducted a retrospective cohort study of individuals who underwent diagnostic sleep studies at the University of Kansas Health System Sleep Disorders Center (2004-2025) and had a 12-lead ECG acquired within one year. Standard ECG features (intervals, amplitudes, voltage-time integrals, morphology descriptors) were extracted from the representative beat. Latent representations were derived using two variational autoencoder (VAE) models pretrained on >1 million ECGs: a 30-dimensional representative-beat VAE and a 90-dimensional full 10-second ECG VAE. Three LightGBM models were trained: clinical factors only (age, sex, weight, height, body mass index [BMI], Epworth sleepiness scale [ESS]), ECG features only, and a combined model. Feature selection and hyperparameter tuning were performed separately for each model within a 5-fold cross-validation scheme. The outcome was moderate-to-severe OSA (AHI[&ge;]15 events/h), evaluated by AUROC (bootstrapped 95% CI) in a 10% holdout test set. ResultsThe dataset included 7,699 individuals (52.2% women, mean age 58.0 {+/-} 14.2 years, BMI 33.6{+/-}8.2 kg/m{superscript 2}, ESS 8.8{+/-}5.1; 36.4% with moderate-to-severe OSA) Training AUROC (95% CI) was 0.71 (0.66-0.76) for the combined model, versus 0.65 (0.59-0.70) for the clinical-only model and 0.66 (0.61-0.72) for the ECG-only model. Holdout test AUROC was 0.78 (combined), 0.69 (clinical-only), and 0.71 (ECG-only). Feature importance identified age, BMI, weight, ECG VAE latent variables, height, sex, and P-wave voltage-time integral as relevant contributor to the predictions. ConclusionsLatent ECG features extracted by VAE improve prediction of moderate-to-severe OSA when combined with clinical factors.

Background Cardiovascular prevention is organized around target metrics. Blood pressure, LDL cholesterol, and A1c define treatment targets, quality metrics, and, implicitly, who is considered safe. Yet a value below a cutoff is a correlative measurement, not an outcome. We tested whether social context recorded before the measurement separated subsequent cardiovascular events among people whose numbers already met the cutoff. Methods In this retrospective cohort study, we analyzed All of Us participants without recorded cardiovascular disease and with low comorbidity in three separately indexed cohorts: systolic blood pressure (SBP), LDL cholesterol, and A1c. A preindex score combined household income, education, food security, physical activity, and ZIP-linked deprivation, standardized against a fixed pooled reference of 325,652 eligible people. The outcome was first myocardial infarction, ischemic stroke, or heart failure. A sensitivity analysis indexed participants at only events recorded in inpatient or emergency settings. Results The SBP, LDL, and A1c cohorts included 19,525, 16,578, and 18,539 participants with 144, 129, and 191 events. Each one-SD more favorable score was associated with lower hazard: HR 0.611 (95% CI 0.498-0.749), 0.631 (0.512-0.778), and 0.544 (0.457-0.648), respectively. Below each study cutoff (SBP <140 mmHg, LDL <100 mg/dL, A1c <8%), less favorable context had event rates of 3.39 versus 1.68, 7.16 versus 2.05, and 8.67 versus 2.34 per 1,000 person-years. In a one-year analysis restricted to 2019-2023 index dates, the corresponding adjusted HRs were 0.530, 0.336, and 0.365. In the sensitivity cohorts (54,338, 41,451, and 33,607 participants; 493, 288, and 316 events), HRs were 0.732 (0.657-0.815), 0.660 (0.573-0.761), and 0.631 (0.552-0.722). Conclusions A biomarker below the study cutoff did not identify a uniform, low-risk population. Social context recorded before the measurement separated cardiovascular events after adjustment for the measured value and clinical factors. Risk assessment that stops at the number misses this variation.

Background: Sleep-disordered breathing (SDB) is an established adverse prognostic factor in heart failure (HF). Sleep apnea-specific hypoxic burden (SASHB), which reflects the duration and depth of respiratory event-related oxygen desaturation, has shown stronger prognostic value than the apnea-hypopnea index (AHI) in general populations and patients with coronary artery disease. However, its clinical utility in patients hospitalized due to acute HF remains unclear. Objectives: This study evaluated the prognostic value of SASHB in patients hospitalized for acute HF and compared its performance with AHI for risk stratification and identification of patients who may benefit from positive airway pressure (PAP) therapy. Methods: This post hoc analysis included 230 patients from a prospective single-center observational cohort of adults hospitalized for acute HF. The primary analysis included 178 patients who did not receive PAP treatment. Associations of SASHB and AHI with the composite endpoint of cardiovascular death or rehospitalization for worsening HF were evaluated using Cox proportional hazards models. Model improvement after adding SASHB or AHI to a model including established clinical risk factors was assessed using changes in the C index and Brier score, reflecting changes in model discrimination and prediction error, respectively. In the full cohort, the interaction between PAP treatment and SDB metrics was evaluated using the g-formula. Results: During a median follow-up of 3.7 years, 58 of 178 patients (32.6%) reached the composite endpoint. Both SASHB and AHI were independently associated with the composite outcome after multivariable adjustment. However, only SASHB consistently predicted outcomes across analyses and improved model performance beyond established prognostic factors (bootstrap {Delta} C index, 0.020 [95% CI: 0.008 - 0.031]; bootstrap {Delta} Brier score, - 0.007 [95% CI: - 0.014 to - 0.001]), whereas AHI did not. In the analysis including PAP-treated patients, both SASHB and AHI showed interaction with PAP treatment. Conclusions: In patients hospitalized for acute HF, SASHB showed a consistent independent association with adverse clinical outcomes and provided additional prognostic information beyond AHI. SASHB may improve risk stratification and may help identify patients who are more likely to benefit from PAP therapy.

Background: Identification of in-hospital cardiac arrest (IHCA) through manual chart abstraction is the gold standard, but its time-consuming and resource-intensive, which naturally limits its practicality. Diagnostic codes provide an accessible automated alternative, but prior work has shown this method of extraction has poor sensitivity and poor positive predictive value. More accurate and scalable automated methods are needed to support IHCA surveillance and research. Methods: We developed large language model (LLM)-based systems that incorporated physician-defined clinical knowledge through guideline-enriched prompting, supervised fine-tuning, and a self-correction procedure. The systems identified IHCA events and their locations from electronic health record notes. Performance was evaluated at the encounter level against physician-adjudicated chart review and compared with ICD-code-based identification. Results: Compared with an ICD-code-based algorithm (precision 0.68, recall 0.91, F1 score 0.78), the supervised fine-tuned LLM offered substantially better performance for IHCA identification on a curated evaluation corpus, reaching a precision of 0.88, a recall of 1.00, and an F1 score of 0.94. When tested on a larger, unselected validation cohort of 45,525 encounters, a separate LLM configuration - relying on guideline-enriched prompting and a self-correction procedure rather than fine-tuning - achieved a precision of 0.79 for encounter-level IHCA identification, meaning most encounters flagged by the model were genuine IHCA cases. The fine-tuned LLM underperformed on this cohort due to its inability to self-correct its decisions, an issue that can be mitigated by having a separate model perform the correction step. Conclusion: Explicit integration of clinical knowledge through prompting and supervised fine-tuning enabled accurate identification of IHCA events and locations from routine clinical documentation. In a two-pass workflow, these systems could prioritize encounters for clinician adjudication, reduce manual screening requirements, and support more scalable IHCA surveillance and registry development.

Rationale: Obstructive sleep apnea (OSA) is highly prevalent yet largely under-diagnosed. Current screening strategies rely on effortful questionnaires with modest accuracy, and prior machine learning models often require resource-intensive inputs and lack external validation. Objectives: To develop and validate machine learning models to predict OSA (apnea-hypopnea index [AHI4%][&ge;]5) and moderate-severe OSA (AHI4%[&ge;]15) using only routinely collected electronic health record (EHR) data, test cross-system transferability and compare performance with clinical screeners and a Bayesian risk-updating framework. Methods: Three tree-ensemble families (XGBoost, random forests, and histogram-based gradient boosting) were trained on an adult cohort with sleep-study results at Kaiser Permanente (KPSC), using full (106-feature) and minimal (age, sex, body-mass index, race/ethnicity) predictor sets and tuned by nested cross-validation. XGBoost models were selected, internally and temporally validated, and externally validated at University of Kansas Medical Center (KUMC). Measurements and Main Results: On our internal hold-out set (n=53,048), the full XGBoost model reached a ROC-AUC of 0.791 for any OSA and 0.768 for moderate-severe OSA. The minimal model retained most performance (0.764 and 0.736). Temporal performance was stable. Externally, frozen models showed lower AUCs (0.718 and 0.690 for full models; 0.725 and 0.706 for minimal models) but similar to KUMC-native models and improved further with transfer learning. The models significantly outperformed STOP-BANG (AUC difference 0.06-0.09 for predicting any OSA; p<0.001). Bayesian updating with select patient-reported items further improved discrimination. Conclusions: Machine learning models based on EHR data can predict OSA probability that outperform existing screeners, transfers across systems, and offers potential for EHR implementation.

The biomechanical properties of the human aorta play a key role in cardiovascular function and disease, yet their population-wide trends and variations remain largely unknown. To fill this crucial gap, we quantified variations in aortic size, stiffness and stresses across 57,434 individuals from the UK Biobank using two-dimensional cine MR imaging and a combined deep-learning, probabilistic and biomechanical modelling framework. Aortic diameter and distensibility (inverse of stiffness) were calculated for the ascending and descending aorta, and their conditional distributions were characterised as a function of age, sex, body size and blood pressure. These observations were integrated with ex-vivo tissue mechanics to infer stress distributions across the aortic wall and the associated uncertainty. Our analysis revealed several new relationships. Larger aortas were found to be stiffer, and greater body size (height and/or weight) was associated with larger aortas with no change in stiffness. Aortic distensibility was negatively associated with systolic, but not diastolic, blood pressure. Apparent sex differences in aortic size were substantially reduced after accounting for body size. Importantly, the maximum stress was found to depend not only on aortic size, as previously thought, but also on distensibility, with lower distensibility associated with higher stresses. The results provide first-of-their-kind population-scale reference distributions for human aortic stiffness, which will improve the personalisation of computational models and digital twins. The correlation between stress and distensibility strongly supports progressing beyond the current diameter-based criteria (based on the law of Laplace) and for the inclusion of stiffness in the clinical risk assessment of aortic diseases. Overall, this study forms the foundation for personalised quantitative medicine and biomechanically informed risk stratification of aortic aneurysms and dissections.

ImportanceCardiovascular disease remains the leading cause of mortality worldwide. Although machine learning models have demonstrated strong predictive performance, limited attention has been given to uncertainty quantification and gender-specific risk factor identification, which are essential for clinically trustworthy artificial intelligence. ObjectiveTo develop and evaluate an interpretable and uncertainty-aware machine learning framework for cardiovascular disease prediction that integrates an association study on biomarkers, explainable artificial intelligence, and gender-stratified risk assessment. Design, Setting, and ParticipantsThis retrospective observational study utilized the UCI Heart Disease dataset comprising 918 participants, including 725 males and 193 females. Demographic, clinical, laboratory, and exercise-related variables were analyzed. Missing values were addressed using a hybrid imputation strategy combining MICE and supervised Random Forest imputation. MethodsAn ensemble machine learning imputation strategy was developed to address missingness. Then advanced machine learning models were developed and evaluated using stratified training/testing partitions and 5-fold cross-validation. Predictive performance was assessed using several evaluation metrics. Uncertainty was quantified through bootstrap resampling, predictive entropy, and mutual information. Linear Double machine learning and Causal Forest double machine learning were applied to estimate gender-specific averages and association studies for risk factors. ResultsRandom Forest achieved the strongest and most stable predictive performance across male and female cohorts, with ROC-AUC values exceeding 0.90 and robust cross-validation performance. Hybrid imputation improved classification performance, increasing the F1-score from 0.87 to 0.89 and ROC-AUC from 0.90 to 0.92. The causal association study analyzes identified the number of major vessels, ST-segment slope, thalassemia, and exercise-induced angina as the strongest determinants of heart disease, with notable differences observed between male and female subgroups. Conclusions and RelevanceIntegrating predictive modeling, uncertainty quantification, and association study machine learning provides a robust and clinically interpretable framework for cardiovascular disease risk assessment. The proposed approach improves prediction reliability while identifying gender-specific causally important risk factors, supporting precision cardiovascular medicine and trustworthy clinical artificial intelligence.

Background: Transthoracic echocardiography (TTE) remains the gold standard for assessing left ventricular (LV) diastolic dysfunction (DD), but noninvasive alternatives are needed to overcome geometric and acoustic limitations. 4D-Flow MRI enables comprehensive assessment of intraventricular pressure differences (IVPDs) and flow propagation velocity (VP). We aimed to evaluate 4D-Flow-derived IVPDs and VP in patients with hypertrophic cardiomyopathy (HCM). Methods: Thirty-eight HCM patients and 38 age- and sex-matched healthy controls underwent TTE and 4D-Flow MRI. Spatiotemporal maps of velocity and IVPDs, decomposed into inertial and convective components, were derived along the main inflow streamline from the mitral valve to the apex. Peak early filling rate (PEFR) was used to normalize IVPDs and VP. Results: While absolute peak flow rates were comparable, HCM patients exhibited significantly faster VP (100.4 vs. 60.2 cm/s, p < 0.01) and higher positive early E IVPD (1.40 vs. 0.92 mmHg, p < 0.01) compared to controls. Normalizing IVPDs by PEFR amplified group differences, revealing a markedly increased flow-normalized E-IVPD (5.13 vs. 2.94 x 103 mmHg.ml1.s, p < 0.001) and deeper reverse ER IVPD (-6.06 vs. -4.61 x 103 mmHg.ml1.s, p < 0.01) in HCM. These differences were primarily driven by the inertial components. Multivariable analysis demonstrated that ER IVPD is independently determined by PEFR and the TTE E/E' ratio (r = 0.65, p < 0.001). Conclusions: 4D-Flow MRI-derived, flow-normalized IVPDs and propagation velocity provide robust, noninvasive markers of altered LV chamber stiffness and impaired diastolic filling in HCM, offering deeper 3D hemodynamic insights into diastolic dysfunction.

Background: Cardiovascular (CV) screening is critical for personnel deploying to Antarctica, where geographic isolation and limited medical care make the management of acute cardiovascular events challenging. The utility of contemporary risk prediction tools such as the AHA PREVENT equations in this unique population remains uncertain. Methods: We retrospectively analyzed 154 individuals referred for enhanced CV evaluation before deployment to the U.S. Antarctic Program (USAP). Participants were evaluated across three temporal cohorts (2018-2020, 2021-2023, and 2024-2026). Demographics, CV risk factors, cardiac testing, and deployment clearance outcomes were collected. Risk was estimated using the Framingham Risk Score (FRS), AHA/ACC Pooled Cohort Equations (PCE), and AHA PREVENT risk models. Results: The cohort was predominantly male (79.9%; mean age 44.6 {+/-} 14.6 years), and abnormal ECG was the most common referral trigger (70.8%). Median 10-year risk was low by PREVENT and FRS and borderline elevated by PCE (PREVENT 2.7% [IQR 1.1-6.6], FRS 2.3% [IQR 0.5-7.1], and PCE 6.1% [IQR 2.4-9.2]). Among 72 participants with all scores, 51% were classified discordantly; PREVENT ASCVD agreed closely with FRS (=0.77) but poorly with PCE (=0.24-0.35), which consistently estimated higher risk. PCE was calculable in 52.6% versus 70.8% for PREVENT. Overall, 132 participants (85.7%) were cleared for deployment, while 8 (5.2%) were not physically qualified (NPQ); 6 of these NPQs were due to cardiovascular causes. Of 22 cardiac medical evacuations during the study period, 21 involved individuals without prior advanced cardiovascular screening. One screened individual was evacuated for symptoms subsequently found to have no cardiac cause. Conclusions: Risk calculators disagree substantially in the fitness-selected population. We recommend PREVENT as the primary risk-stratification tool as it is the most contemporary derivation, incorporates renal function, is race-agnostic, extends to heart failure, applies from age 30, and agrees closely with FRS, easing transition. Targeted imaging, including coronary artery calcium scoring, may supplement intermediate-risk candidates.