European Heart Journal - Digital Health

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: 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.

Rural patients with heart failure (HF) have poor access to care, worse outcomes, and may benefit from at-home monitoring. We evaluated whether foot temperature measured using a thermographic floor mat was associated with invasively measured cardiac output and whether longitudinal temperature changes preceded HF events. We conducted a two-part prospective, observational single-center study using the SmartMat (Podimetrics, Somerville, MA), which acquires plantar thermograms and ambient temperature. In the acute study, patients undergoing clinically indicated right heart catheterization completed 4 scans on the same day. In the chronic study, patients with a HF hospitalization in the past year and New York Heart Association Class II-III symptoms were instructed to use the device and a scale daily for 6 months. A HF event was defined as receiving at least one dose of intravenous or subcutaneous diuretic. Thirty patients were enrolled in the acute study. Mean right hallux temperature positively correlated with thermodilution cardiac output (r=0.63, p<0.001). 42 patients were enrolled in the chronic study. Among patients experiencing an HF event, ambient-adjusted hallux temperature declined in the 15 days immediately preceding the event compared with days 15-30 before the event. Plantar thermograms are associated with invasively measured cardiac output and may detect subclinical worsening before HF events. Foot temperature may provide complementary information regarding HF decompensation, supporting further evaluation of plantar thermography for remote HF monitoring.

Purpose To evaluate the external transportability of source-derived probability calibration for chest radiograph deep learning classifiers under distribution shift and determine its dependence on sample size and disease prevalence. Materials and Methods This retrospective secondary computational analysis of public, de-identified radiographic datasets involved no participant contact or new data collection. Cohort sizes were prespecified by availability without prospective power calculation: 106,686 NIH ChestX-ray14 images (29,323 patients; 1992-2015) for development and internal testing, 20,000 CheXpert images (20,000 patients; October 2002-July 2017), and 14,998 VinBigData images (2018-2020) for external testing. ResNet-18, DenseNet-121, and DeiT-Tiny classifiers for cardiomegaly and pleural effusion were trained across four patient fractions and three seeds (36 fits). Calibrators were fitted exclusively on NIH data. Twelve primary external Brier-score contrasts (temperature-scaled versus raw probabilities) were evaluated using paired cluster-bootstrapping with Holm adjustment. Results Temperature scaling reduced Brier scores in nine of 12 primary contrasts but worsened pleural effusion calibration on VinBigData across all architectures (Brier increases: ResNet-18, 0.001460 [95% CI: 0.001347, 0.001568]; DenseNet-121, 0.001193 [95% CI: 0.001109, 0.001274]; DeiT-Tiny, 0.000498 [95% CI: 0.000417, 0.000577]; all adjusted P=.006). Standardizing class weights to NIH prevalence reversed all CheXpert improvements. Although full versus 10% training improved external AUROC across all comparisons, DeiT-Tiny CheXpert pleural effusion Brier score worsened by 0.007039 despite a 0.032466 AUROC gain. At 80% selective image retention, calibrated ensembles retained only 12.3%-17.2% of VinBigData positive pleural effusion examinations. Conclusion Source-fitted calibration is not uniformly transportable across health systems, and discrimination gains do not guarantee calibrated probabilities. External evaluation must simultaneously assess probability reliability, prevalence shifts, and retained-case composition.

Background: Coronary perivascular fat attenuation index score (FAI-Score) on coronary computed tomography angiography (CCTA) quantifies vascular inflammation and predicts cardiac mortality in symptomatic cohorts. However, its prevalence, anatomical distribution, and clinical correlates in asymptomatic primary prevention populations remain uncharacterized. Methods: In an IRB-approved prospective study across two clinical sites, 84 consecutive asymptomatic adults (mean age 53.7 {+/-} 8.6 years; 63.1% men) with no prior coronary artery disease underwent screening CCTA and comprehensive clinical, metabolic, and biomarker profiling. AI-standardized FAI-Score was quantified across three epicardial vessels (LAD, LCx, RCA). High and very high perivascular inflammation were pre-specified as any vessel [&ge;]75th and [&ge;]90th percentiles, respectively. Luminal stenosis and plaque were graded using CAD-RADS 2.0. Results: Elevated FAI-Score ([&ge;]75th percentile) was present in 42.9% of participants (36/84; 95% CI 32.8% - 53.5%), with 17.9% (15/84; 95% CI 11.1% - 27.4%) exhibiting very high inflammation ([&ge;]90th percentile). Multi-vessel inflammation was identified in 17.9% (15/84), with the RCA most frequently involved ([&ge;]90th percentile in 13.1%). Standard circulating biomarkers, including ApoB, non-HDL-C, HbA1c, and PREVENT-ASCVD 10-year risk showed no association with binary inflammation. Systemic hs-CRP (median 0.8 mg/L) exhibited an inverse correlation with maximum vessel FAI-Score (Spearman {rho}=-0.27, p=0.014). Men demonstrated higher rates of elevated FAI-Score than women (52.8% vs. 25.8%, p=0.016; [&ge;]90th percentile: 26.4% vs. 3.2%, p=0.007). Strikingly, 46.2% (30/65) of participants with normal luminal anatomy (CAD-RADS 0) and 47.2% (25/53) with zero plaque burden harbored elevated FAI-Score; all 15 individuals with very high FAI-Score had CAD-RADS 0 (Fisher's exact p=0.018). Body mass index was the sole secondary correlate surviving false discovery rate correction (AUC = 0.71, p=0.001, all-test FDR q=0.028). Regularized multivariable models showed modest out-of-sample discrimination for FAI-Score elevation (cross-validated AUC = 0.71). Conclusions: Coronary perivascular inflammation is common among asymptomatic adults and frequently uncouples from structural plaque and luminal stenosis. Routine circulating biomarkers and clinical risk calculators fail to identify inflamed vessels, highlighting FAI-Score as an independent dimension of primary cardiovascular risk assessment.

Background: The systemic immune-inflammation index (SII) is a routinely available blood-derived marker with potential prognostic relevance after coronary artery bypass grafting (CABG). We evaluated associations between preoperative SII and postoperative outcomes. Methods: PubMed, Embase, Scopus and Web of Science were searched to 1 October 2026, with supplementary searches through 14 October 2026. Peer-reviewed observational studies evaluating preoperative SII in adults undergoing CABG were eligible. Methodological quality was assessed using the Newcastle-Ottawa Scale. Compatible categorical associations and exploratory adjusted continuous mortality effects were pooled separately using restricted maximum-likelihood random-effects models; other findings were summarized descriptively. Results: Thirty-one reports were included, with an aggregate sample of 30,976 and possible cohort overlap; five contributed to meta-analysis. Two off-pump cohorts comprising 1,378 patients and 180 atrial-fibrillation events yielded an imprecise crude association with postoperative atrial fibrillation (OR 1.60, 95%CI 0.53-4.82; I2=68%). An exploratory crude synthesis indicated greater intra-aortic balloon pump use with higher SII (OR 2.45, 95%CI 1.16-5.18; I2=51%). Two cohorts with differing follow-up yielded an exploratory adjusted mortality HR of 1.11 (95% Wald CI 1.05-1.18) per study-specific SD increase in log-SII. In one 624-patient cohort, high SII was associated with postoperative acute kidney injury after adjustment (OR 3.35, 95% CI 2.25-4.97).Apparent atrial-fibrillation discrimination varied (AUC 0.587-0.946). Conclusions: Preoperative SII showed associations with selected adverse outcomes after CABG, supporting further evaluation as an accessible prognostic marker. Its incremental value beyond established clinical risk models requires validation. Keywords: Systemic immune-inflammation index; Coronary artery bypass grafting; Atrial fibrillation; Prognosis; Systematic review

Background Cardiac amyloidosis is an increasingly recognized cause of heart failure in older adults, yet Colombian data, and data from the southwest of the country in particular, remain scarce. We describe the clinical, diagnostic, imaging and outcome profile of patients with cardiac amyloidosis at two high-complexity institutions in Cali. Methods Observational, descriptive, retrospective, two-center study. Patients were identified in the institutional registries of both centers using ICD-10 codes for amyloidosis, and clinical data were abstracted manually from medical records. Continuous variables are reported as median (interquartile range) and categorical variables as counts and percentages over the number of patients with available data. Results Thirty-two patients were analyzed (median age 77 years; 68.8% men). Heart failure (90.6%), atrial fibrillation (46.9%) and carpal tunnel syndrome (31.2%) were frequent. The etiological classification depended on the test considered: bone scintigraphy was positive (Perugini 2-3) in 23 of 26 patients studied (88.5%), but a monoclonal protein was screened for in only 22, and the full non-invasive criteria for ATTR were met in 16 of 32 (50.0%). Genetic testing was positive in 9 of 18 tested, and the variant was Val142Ile in all 8 patients in whom it was characterized. The working diagnosis was ATTR in 26 patients (81.2%) and AL in 6 (18.8%). Tafamidis was used in 37.5%. Eleven of the 31 patients with known vital status died (35.5%), and NT-proBNP was markedly higher among those who died (6640 vs 1186 pg/mL; p=0.003). Conclusions In this southwestern Colombian cohort, cardiac amyloidosis presented as the classic transthyretin phenotype, with the Val142Ile variant in all 8 patients in whom it was characterized. Bone scintigraphy was frequently positive, but screening for a monoclonal protein was often omitted, so only half the cohort met the full non-invasive criteria for ATTR. Mortality was high and closely associated with NT-proBNP.

Background: Anatomic isthmus 3 (AI 3) is defined as the myocardium between the pulmonary valve annulus and the ventricular septal defect patch that persists after tetralogy of Fallot (ToF) repair. Ventricular tachycardia (VT) propagated through slow conducting AI 3 (SCAI 3) represents a significant cause of morbidity in ToF. There is an unmet need for non-invasive identification of SCAI 3 in patients with ToF. Objectives: To assess whether the anatomic features of AI 3 as derived from cardiac computed tomography (CT) can predict slow conduction velocity during EP study. Methods: This retrospective single-center study included patients with ToF who underwent an electrophysiologic study and had AI 3 width measured from the pre-procedure cardiac CT. All patients underwent programmed ventricular stimulation, and electroanatomic. Logistic regression was performed to assess the relationship between AI 3 width and the presence of SCAI 3. Results: Of 27 consecutive patients with ToF, twelve patients were found to have SCAI 3, and ten patients were found to have SCAI 3 mediated VT. The AI 3 was significantly narrower in patients with SCAI 3 compared to those with normal AI 3 conduction (18.4 +/- 6.6 mm vs. 29.2 +/- 9.0 mm; p = 0.001). An AI 3 width above 21.5 mm carried a negative predictive value of 80.0%. Conclusion: In adults with repaired ToF referred for electrophysiology study, a narrower CT-derived AI 3 width was associated with SCAI 3 identification. AI 3 width measured on a CT obtained before PV replacement could help guide invasive risk stratification.

Patients with implantable cardioverter-defibrillators (ICDs) often are excluded from receiving CMR scans at 3.0T due to ICD-induced image artifacts. This work aims to transfer sequence protocols established for CMR imaging in the presence of ICDs at 1.5T to a first human application at 3.0T. All measurements were performed on a clinical 3T MRI scanner. Phantom measurements were used to establish and optimize wideband PSIR LGE imaging at 3.0T. Volunteer measurements with externally placed ICDs were used for optimization of CINE imaging and wideband PSIR imaging. Thirdly, patients with implanted ICD were examined with the optimized protocols. Human measurements showed that SA stacks obtained with spoiled GRE after contrast agent showed a markedly lower artifact burden than the conventional approach. For scar imaging, a wideband PSIR with a bandwidth of 3kHz and a frequency offset of +750Hz allowed a sufficient suppression of relevant artifacts in the majority of scans. No strong correlation between the image quality and the distance between the cardiac silhouette and the device was detected. The presented work shows the feasibility of clinical cardiac MRI at 3.0T including GRE-CINE and wideband PSIR-LGE for patients with ICDs.

Background: Conventional performance metrics do not indicate whether artificial intelligence (AI)-platforms and radiologists are right or wrong on the same examinations, nor whether their non-agreement is more often beneficial or harmful. Objective: To develop, illustrate, and apply directional metrics quantifying non-agreement between radiologists and AI-platforms against a common reference standard, and to adapt a fragility-based framework for quantifying the statistical robustness of these metrics. Methods: We defined Positive Non-Agreement (PNA, AI-platform correct, radiologist incorrect), Negative Non-Agreement (NNA, radiologist correct, AI-platform incorrect), and Net Non-Agreement (NENA, their difference). Metrics were illustrated in three hypothetical scenarios of 1,000 examinations each and applied to three previously published AI-radiologist studies on pulmonary embolism and mammography. Bootstrap confidence intervals were calculated, and a reclassification fragility index (FI) quantified the minimum case reclassification needed to remove statistical support for NENA's direction. Results: In illustration, NENA was positive in two scenarios and negative in one. In the three published studies, NENA ranged from -4.18% to +10.74% and decreased consistently with increasing radiologist experience (junior residents, +6.27%; consultants, +1.45%), a gradient driven primarily by declining PNA with modest increase in NNA. The reclassification FI was defined for all three illustrative scenarios (range, 0.10%-1.70%) and for six of the seven comparisons from these studies (range, 0.34%-4.31%), but not for the weakest comparison, whose confidence interval for NENA included zero. Conclusions: PNA, NNA, and NENA provide a directional, case-level complement to conventional performance metrics for AI-radiologist non-agreement. The accompanying reclassification FI offers a statistically grounded, generalizable method for quantifying their robustness.

BACKGROUND Timely cause-of-death (COD) information guides health policy, yet many deaths in low- and middle-income countries are not medically certified. In Africa, fewer than 10% of deaths are. Verbal autopsy (VA) yields a structured symptom questionnaire and a free-text narrative. Mortality surveillance increasingly records the narrative alongside the questionnaire, but the algorithms used to produce VA-based statistics were designed for structured responses only, so narratives inform physician COD assignment but do not contribute to algorithmic COD assignment. We evaluated whether supervised models could use both modalities to improve COD assignment. METHODS We analyzed 4,999 VA records for deaths at ages 12 years or older from a national sample of registered deaths in South Africa (RSA), using physician-assigned CODs as the reference standard. We trained unimodal and multimodal pipelines on the labelled records: pretrained language models (PLMs) fine-tuned on narratives alone, machine learning (ML) models on structured responses alone, and the two combined at the data, feature, and decision levels. We also prompted a locally deployed generative large language model (LLM) on the same unimodal and multimodal inputs, without parameter updates. All were compared with InSilicoVA, the operational comparator representing current automated practice. We evaluated individual-level assignment, population-level agreement, calibration, and generalization across provinces in RSA, reporting these metrics rather than ranking pipelines on any one of them. RESULTS Every supervised pipeline improved on InSilicoVA by 8.5-15.9 percentage points in top-1 accuracy and 3.5-7.1 points in cause-specific mortality fraction (CSMF) accuracy. The narrative and structured pipelines performed similarly in aggregate but differed by cause: 63.9% of deaths received identical predictions, 11.6% were correct only from the narrative pipeline, and 13.6% only from the questionnaire pipeline, with the narrative pipeline more accurate for noncommunicable causes and the questionnaire pipeline for communicable causes. Against the unimodal approaches, the decision-fusion pipelines added 4.2-5.4 percentage points in top-1 accuracy, concentrated in the most common causes, whereas feature fusion with cross-modal attention improved the less common causes, adding 6.6 points (95% CI, 1.7-11.5) in balanced accuracy and 5.5 points (95% CI, 0.6-10.3) in macro F1. Feature fusion reached the highest CSMF accuracy (95.7%). Paired differences from the better unimodal representative ranged from -0.2 to +3.4 points, with all intervals including zero. The LLM exceeded InSilicoVA on individual-level metrics and reached the highest balanced accuracy of any pipeline (60.6%), while the supervised pipelines retained higher CSMF accuracy. Supervised pipelines had lower expected calibration error than InSilicoVA. Supervised gains held across all nine held-out provinces, including every top-1 comparison and nearly every macro F1 and CSMF comparison. CONCLUSIONS PLMs using narratives alone substantially improved on current best practice and matched ML pipelines using structured responses, while multiple multimodal strategies produced further gains in individual COD assignment. The zero-shot LLM showed particular strength on rare causes, whereas supervised pipelines better recovered the population cause distribution. These findings support incorporating narratives as a core input to automated VA systems and evaluating representative fusion strategies for the intended output. External validation across settings and age groups will support broader implementation.

Study Objectives: Glucagon-like peptide-1 (GLP-1) receptor agonists raise resting heart rate. We asked whether, after semaglutide initiation, the increase is larger during rapid eye movement (REM) sleep than during non-rapid eye movement (NREM) sleep. Methods: In the Human Phenotype Project, an Israeli cohort of generally healthy adults aged 40 to 70 years, 96 of 494 self-reported semaglutide initiators had a home sleep study (WatchPAT; sleep stages estimated from peripheral arterial tone and actigraphy) before and at least three months after starting. Each was propensity-matched 1:1 to a never-user of GLP-1 drugs measured at the same visits. We estimated the initiators' change minus their controls' change (difference-in-differences). The REM minus NREM contrast was designated primary after an exploratory analysis. Results: Relative to controls (90 to 94 pairs), heart rate rose during nocturnal wake (+3.89 beats per minute [bpm], 95% confidence interval 2.35 to 5.44), NREM sleep (+4.82 bpm, 3.12 to 6.52) and REM sleep (+6.25 bpm, 4.42 to 7.91); the REM increase exceeded the NREM increase by 1.42 bpm (0.98 to 1.89). The excess persisted across matching designs and on a relative scale; the proportion of REM sleep did not change. Never-users with similar weight loss showed no REM sleep heart-rate rise (+0.40 bpm, -0.63 to 1.50). Heart-rate variability decreases disappeared after adjustment for heart rate. Conclusions: Semaglutide initiation was associated with higher heart rate throughout sleep, most during REM sleep, unexplained by weight loss. Clinic measurements may understate this effect. Confirmation with polysomnographic staging is needed.

Background. Immune checkpoint inhibitors (ICIs) can cause immune-related adverse events in almost any organ. While myocarditis has received the most attention, data on pericardial disease with ICI use are limited. We examined pericardial adverse event reporting with nine ICIs using the FDA Adverse Event Reporting System (FAERS). Methods. We combined legacy AERS and current FAERS quarterly files from January 2011 to June 2026 into 18,629,240 unique cases, deduplicated by case number, and removed cases the FDA had listed as deleted. ICI exposure was identified from generic names, brand names, development code names and the active-ingredient field. Pericardial events were defined by seven MedDRA preferred terms. We estimated reporting odds ratios (ROR), proportional reporting ratios (PRR), Bayesian information components (IC) and empirical Bayes geometric means (EBGM). We also examined myocarditis co-reporting, time trends, a comparison with other antineoplastic agents, a MedDRA query-based outcome, and 14 alternative definitions. Results. The analysis included 18,514,045 cases: 277,434 involved an ICI, 34,792 carried a pericardial term, and 1,680 had both. Pericardial events were reported disproportionately with ICIs (ROR 3.35 (95% CI 3.19 to 3.52); PRR 3.34; IC025 1.62). Eight of nine agents had a ROR above 1 (3.32 to 5.63). Among the most frequently reported terms, estimates were highest for cardiac tamponade (5.59 (95% CI 4.98 to 6.28)), followed by pericardial effusion (3.83 (95% CI 3.60 to 4.07)) and pericarditis (2.09 (95% CI 1.89 to 2.31)). After removing cases that also reported myocarditis, the ROR was 3.13 (95% CI 2.98 to 3.30). Most sensitivity analyses gave similar estimates (ROR 3.13 to 3.47); defining exposure by generic drug names alone, which identified only 44% of exposed cases, gave 3.99. Compared with cases naming other L01 antineoplastic agents, the ROR was 1.13 (95% CI 1.08 to 1.19): 1.28 (95% CI 1.20 to 1.36) for ICI-only cases and 0.96 (95% CI 0.89 to 1.04) when another antineoplastic was also named. The ICI-specific reporting odds ratio fell each quarter (odds ratio 0.977 per quarter, 95% CI 0.973 to 0.980). Conclusions. In FAERS, pericardial events were reported more often with ICIs than with other drugs across most agents and terms; however, incidence and causation cannot be determined. The excess over other antineoplastic agents was small, and reporting disproportionality declined over time. These findings support clinical awareness of pericardial disease in patients receiving ICIs and warrant study in data that can measure incidence.

BACKGROUND: The chromosome 9p21.3 locus is among the strongest genetic risk loci for cardiovascular disease (CVD), yet the regulatory mechanisms underlying this association remain poorly understood. Previous vascular disease studies focused predominantly on the role of this locus in smooth muscle cell (SMC) dysfunction, whereas its potential role in endothelial cells (ECs) remains less well characterized. Here, we sought to identify candidate regulatory variants at 9p21.3, their distal target genes, and effector transcripts that contribute to CVD-related phenotypes in ECs. METHODS: Promoter capture Hi-C was used to map chromatin interactions at 9p21.3 in primary human ascending aortic SMCs (n=6), supplementing the previously obtained ECs (n=16) dataset. To support the regulatory relevance of these interactions, expression quantitative trait locus analysis was performed in 136 vascular biopsies. Variant effects were prioritized in silico using AlphaGenome. The regulatory activity of the prioritized region was validated using CRISPR activation. RESULTS: Chromatin interaction mapping identified a complex regulatory architecture at the 9p21.3 locus involving 19 putative regulatory elements and 15 genes across ECs and SMCs. MIR31HG showed the most recurrent distal promoter connectivity with the 9p21.3 region. Consistent with the chromatin interaction map, expression of MIR31HG transcripts was associated with genetic variation at the 9p21.3 locus. In ECs, the lncRNA MIR31HG, but not miR-31, was responsive to hemodynamic flow. Depletion of MIR31HG in ECs was associated with endothelial activation. Epigenetic tracks and deep learning predictions highlighted an active enhancer element at 9p21.3, which was subsequently validated by CRISPR activation and increased MIR31HG expression, confirming its capacity to regulate MIR31HG over a large genomic distance of approximately 538 kb. CONCLUSIONS: Altogether, these data implicate lncRNA MIR31HG as a likely effector of the 9p21.3 CVD locus, linking the cis-regulatory element and heritable genetic variation in the respective region to endothelial activation through long-range enhancer regulation.

Background: Hospital readmission is a common and costly outcome in patients with chronic obstructive pulmonary disease (COPD). Multimodal deep learning models combining structured electronic health record (EHR) data with physiological, imaging, and textual data have been proposed to improve its prediction, but whether such fusion outperforms EHR data alone, once temporal-leakage safeguards are applied, remains unclear. Methods: Using a stratified MIMIC-IV cohort of COPD patients enriched to about 50% readmission (natural prevalence 18.1%; n=1,279; 895/192/192 training/validation/test), we predicted 30-day, all-cause readmission from four frozen, pre-trained encoders (electrocardiogram, chest radiograph, clinical notes, structured EHR). We audited and corrected temporal leakage in the admission-linkage pipeline of each modality, then compared three fusion architectures (naive concatenation, scalar gated fusion, EHR-anchored residual fusion) against unimodal baselines and an EHR-alone reference, using bootstrap significance testing, a missingness stress test, and calibration analysis with and without post-hoc recalibration. Results: The audit corrected three previously undetected temporal-alignment issues, affecting 5-9% of patients for clinical notes, 29-43% for chest radiographs, and 53-58% for ECG. After correction, every modality except EHR showed near-chance discrimination alone (ROC-AUC 0.46-0.53), confirmed with a non-linear adapter probe. No fusion architecture significantly outperformed the EHR-alone baseline (ROC-AUC 0.63); removing auxiliary modalities improved rather than degraded fusion performance under stress testing. The raw Brier score was lower for all three fusion architectures than for EHR-alone (about 0.24 versus 0.32), but this reflected overconfidence of the EHR-alone probe (calibration slope 0.12); after logistic recalibration fitted on the validation set, Brier scores were indistinguishable (EHR-alone 0.239; fusion 0.240-0.244; all paired bootstrap 95% CIs included zero). Conclusions: Structured EHR data alone matched the discrimination and, after recalibration, the calibration of more complex multimodal fusion for 30-day COPD readmission prediction. These findings highlight the importance of rigorous temporal-leakage auditing in multimodal clinical prediction research and suggest that reformulating this task around temporal proximity to the outcome, rather than refining the fusion mechanism further, may be more productive for future work.

INTRODUCTION: The digital clock drawing test (dCDT) is a cognitive screening tool capturing high-resolution pen-movement data. Existing approaches for classifying cognitive impairment from dCDTs rely on selected features or image-based methods. We propose representing dCDTs as structural networks to classify cognitive impairment and characterize spatial organization in clock drawings. METHODS: Weighted networks were constructed from 3,362 Framingham Heart Study participants, with and without cognitive impairment. Network representations were generated separately for command and copy conditions, with nodes representing pre-defined clock components (digits, hands, clock face) and edges capturing pairwise spatial relationships between components. Random forest models and graph convolutional networks were trained to predict cognitive impairment from the networks. Models were benchmarked against demographics-only and established summary feature baselines. RESULTS: Network-based models achieved AUC values comparable to the established summary feature baseline across command-only and copy-only conditions. Differences in spatial organization between the cognitively intact and impaired groups were greatest in the upper-right clock region, involving digits 1 and 2, the clock hands, and the surrounding clock face. Differences between the command and copy networks were larger among cognitively impaired participants. DISCUSSION: The proposed network representations provide an objective, scalable, and flexible framework for representing clock drawings for use in classification models. dCDT networks matched the predictive performance of established summary feature-based approaches and identified clock regions where spatial organization differed by cognitive status. After external validation, this approach has the potential to be incorporated within a clinical workflow to support automated and interpretable cognitive screening.

Rationale: Conventional CT-derived pulmonary vascular metrics quantify vascular pruning but do not assess the spatial heterogeneity of vascular architecture. In COPD, peripheral microvascular pruning coexists with compensatory central vessel dilation, paradoxically preserving total vessel volume while reducing effective vascular penetration into lung parenchyma. Objectives: To develop and evaluate Vascular Proximity Coverage (VPC), a distance-based CT biomarker quantifying the spatial efficiency of pulmonary vascular distribution. Methods: This retrospective analysis included 7,881 COPDGene participants [PHASE] with complete vascular segmentations. VPC was estimated by integrating the area under the cumulative lung tissue coverage curve at distances of 0-10 mm from segmented vessel voxels. Multivariable models assessed associations with lung function, six-minute walk distance (6MWD), St. George's Respiratory Questionnaire (SGRQ) score, mMRC dyspnea score, and all-cause mortality over a median follow-up of [10.8] years, adjusting for demographics, smoking, emphysema, scanner type, and BV5. The cohort was stratified at median BV5 and VPC into concordant (high/high, low/low) and discordant (high-BV5/low-VPC, low-BV5/high-VPC) vascular phenotypes. Results: VPC declined across disease severity (5.82 {+/-} 0.50 [UNITS] in non-smokers to 5.21 {+/-} 0.64 in GOLD 4; p < 0.001). Among spirometrically normal smokers, current smokers had lower VPC than former smokers (5.31 vs. 5.68; p < 0.001). Each standard deviation decrease in VPC was independently associated with reduced 6MWD ({beta} = -16.8 m, 95% CI: -19.5 to -14.0), worse SGRQ ({beta} = 1.9, 95% CI: 1.4 to 2.4), and greater dyspnea (RR = 1.05, 95% CI: 1.03 to 1.08; all p < 0.001). The lowest VPC quartile was associated with 38% higher mortality than the reference quartile (HR = 1.38, 95% CI: 1.18 to 1.62; p < 0.001). The high-BV5/low-VPC phenotype showed 24% higher mortality than high-BV5/high-VPC (HR = 1.24, 95% CI: 1.08 to 1.42; p = 0.002) despite preserved vessel volume. Conclusions: VPC captures spatial heterogeneity of vascular architecture independent of vessel volume, is reduced in smokers without airflow obstruction, and identifies excess mortality risk among individuals whose total vessel volume appears preserved.