European heart journal. Digital health最新文献

{"title":"Health Electronic Assessment of Risks and Trends using Biometric Equipment and Technology (HEARTBEAT): study design and rationale.","authors":"Nassir Marrouche, Christian Massad, Han Feng, Ala Assaf, Mayana Bsoul, Yara Menassa, Ghassan Bidaoui, Qussay Marashly, Chanho Lim","doi":"10.1093/ehjdh/ztag132","DOIUrl":"10.1093/ehjdh/ztag132","url":null,"abstract":"<p><strong>Aims: </strong>Cardiovascular disease remains the leading driver of morbidity and mortality worldwide. As consumer smartwatches become ubiquitous, they offer a practical platform for continuous, real-world phenotyping, capturing passive and active biometrics that may detect occult disease and anticipate adverse cardiovascular events. Scaled deployment of these devices could enable remote screening, longitudinal monitoring, and earlier risk mitigation beyond traditional episodic care.</p><p><strong>Methods and results: </strong>The Health Electronic Assessment of Risks and Trends using Biometric Equipment and Technology (HEARTBEAT) study is a prospective, single-arm wearable study that has enrolled 897 participants since 10 October 2024. The primary objectives are to: (1) define associations between smartwatch-derived biometrics and incident cardiovascular events; (2) derive biometric profiles that support identification of underlying cardiometabolic and cardiovascular conditions; and (3) apply artificial intelligence to improve prediction of cardiovascular events. Participants are screened, consented, and enrolled in person or remotely. Data are captured through electronic medical record (EMR) extraction and a companion smartphone application. Participants are followed for 12 months, with outcomes adjudicated on an ongoing basis using EMR review and app-based surveys.</p><p><strong>Conclusion: </strong>The HEARTBEAT study will test whether scalable, consumer-grade wearables can move from wellness tracking to clinically meaningful signal, identifying comorbidities and predicting adverse cardiovascular outcomes in routine care settings. If successful, the study will help establish an evidence base for smartwatches as validated digital health tools for remote screening and continuous cardiovascular monitoring.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag132"},"PeriodicalIF":4.4,"publicationDate":"2026-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528244/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868234","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A machine learning approach to conglomerate multi-domain features of cardiac aging.","authors":"Glades H M Tan, Enyu Yang, Bryan Z Y Tan, Hane Naghshbandi, Johnathan Loh, Xinliu Zhong, Jun Liu, Daniel J Lim, Fei Gao, Jean-Paul Kovalik, Ru-San Tan, Si Yong Yeo, Angela S Koh","doi":"10.1093/ehjdh/ztag136","DOIUrl":"10.1093/ehjdh/ztag136","url":null,"abstract":"<p><strong>Aims: </strong>Owing to the breadth of complex and highly dimensional clinical data associated with ageing, integration of multiple health domains is needed towards determining cardiac outcomes of older adults. We designed a machine learning (ML) approach to conglomerate multi-domain data and identify determinants of cardiac function in older adults.</p><p><strong>Methods and results: </strong>We applied a structured ML pipeline including data pre-processing, feature selection, and model development using Random Forest, Gradient Boosting, XGBoost, LightGBM, and support vector machine. Model performance was evaluated using stratified k-fold cross-validation and complementary discrimination metrics, including ROC-AUC, PR-AUC, balanced accuracy, sensitivity, and specificity. Feature importance was assessed using Random Forest (RF) importance and Shapley Additive exPlanations (SHAP), and the Tree-based Pipeline Optimization Tool (TPOT) was used for model optimization. The outcome was an impaired myocardial relaxation phenotype based on the mitral peak early-to-late diastolic filling velocity (E/A) ratio. The multi-domain dataset included demographic characteristics, clinical risk factors, physical activity, body composition, serum biomarkers, omics, and cardiac imaging, comprising 227 features from 984 older adults. Thirty key features were identified, mainly related to physical function and metabolomics. Using these features, the selected classifiers achieved ROC-AUC values above 0.79. XGBoost was retained as the primary tree-ensemble benchmark, with cross-validated ROC-AUC 0.8157 and test-set ROC-AUC 0.7658; TPOT was comparable (test-set ROC-AUC 0.7675). Higher XGBoost score was associated with death-or-admission events (HR 1.115, <i>P</i> = 0.029).</p><p><strong>Conclusion: </strong>Multi-domain ML identified clinically interpretable signals associated with impaired myocardial relaxation in ageing and with clinical events.</p><p><strong>Trial registration: </strong>ClinicalTrials.gov Identifier: NCT02791139.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag136"},"PeriodicalIF":4.4,"publicationDate":"2026-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13505868/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148820439","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Detection of obstructive coronary artery disease using a deep learning and machine learning ensemble: a retrospective feasibility study.","authors":"Doyoung Park, Linxuan Yan, Arman Ahmad Khan, Wilbert Hsien Hao Ho, Choon Ta Ng, Jonathan Jiunn Liang Yap, Swee Yaw Tan, Khung Keong Yeo, Lohendran Baskaran","doi":"10.1093/ehjdh/ztag134","DOIUrl":"10.1093/ehjdh/ztag134","url":null,"abstract":"<p><strong>Aims: </strong>Early identification of obstructive coronary artery disease (ObCAD) is crucial because it is strongly associated with acute myocardial infarction. We developed a weighted average ensemble model integrating deep learning (DL) and machine learning (ML) to leverage imaging and clinical data for enhancing the detection of ObCAD.</p><p><strong>Methods and results: </strong>A retrospective cohort of 1054 patients was used to develop an ensemble model combining a 3D Vision Transformer with eXtreme Gradient Boosting and CatBoost for binary classification of ObCAD (>50% stenosis). Unstructured data comprised 3D cardiac non-contrast computed tomography (CT) scans, while structured data included 11 demographic and clinical features. Obstructive coronary artery disease labels were derived from corresponding coronary CT angiography. Model performance was evaluated using 10-fold cross-validation with fold-wise Wilcoxon signed-rank testing. The ensemble model achieved a mean receiver operating characteristic area under the curve (ROC AUC) of 0.81 ± 0.04 and an accuracy of 0.76 ± 0.04. It demonstrated a statistically significantly higher ROC AUC than individual component models. Feature importance analysis identified age, chest pain, and sex as the most influential predictors of ObCAD. Gradient-weighted class activation mapping visualization indicated that the 3D Vision Transformer primarily focused on cardiac regions containing coronary artery calcium deposits.</p><p><strong>Conclusion: </strong>Integrating DL-based imaging analysis with ML-based clinical modelling enhances the discriminative performance for ObCAD detection with complementary interpretability. This ensemble framework demonstrates potential to support clinical decision-making by identifying high-risk patients using routine cardiac CT combined with patient-level clinical data. Future studies using external validation and coronary artery calcium scores may further improve risk prediction.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag134"},"PeriodicalIF":4.4,"publicationDate":"2026-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528245/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868255","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hypertension screening via awake-sleep differences in photoplethysmogram signals.","authors":"Jingyuan Hong, Manasi Nandi, Yali Zheng, Jordi Alastruey","doi":"10.1093/ehjdh/ztag131","DOIUrl":"10.1093/ehjdh/ztag131","url":null,"abstract":"<p><strong>Aims: </strong>Hypertension is a major risk factor for cardiovascular diseases. This study proposes a novel hypertension community-based screening framework based on intra-subject awake-sleep differences in photoplethysmography (PPG) indices, using machine learning. We hypothesized that normotensive individuals exhibit greater PPG variation between awake and sleep states than unmanaged hypertensive individuals.</p><p><strong>Methods and results: </strong>The Aurora-BP dataset (<i>n</i> = 180; 138 normotensive, 42 hypertensive) was used for model development, with 18 subjects reserved for internal testing. External validation was performed using the independent CUHK-BP dataset (<i>n</i> = 26; 11 normotensive, 15 hypertensive). Twenty PPG-based indices were extracted, and subject-level <i>P</i>-values from Mann-Whitney <i>U</i>-tests comparing awake and sleep periods were used as model features. Discretized <i>P</i>-values served as inputs for four machine learning models. The support vector machine (SVM) achieved the highest performance, with 81.1 ± 8.4% accuracy and 82.8 ± 8.1% F1-score on the internal test set using all indices. On the external test set, the SVM using only temporal indices achieved 84.6% accuracy and 86.7% F1 score. Temporal indices, especially those linked to the dicrotic notch, showed strong generalizability across datasets.</p><p><strong>Conclusion: </strong>The study demonstrates the feasibility of awake-sleep PPG analysis for hypertension screening, highlighting the potential of wearable PPG devices for ambulatory monitoring.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag131"},"PeriodicalIF":4.4,"publicationDate":"2026-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13520938/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148842334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and validation of an artificial intelligence-based tool to detect subclinical atheroesclerosis using non-mydriatic retinal funduscopic images: the Aitheroscope project.","authors":"Juan Torres-Macho, María Ángeles Sánchez-Uriz, Vyara Hrystova, Jesús Prada-Alonso, Jose Manuel López-Aragonés, Nuria Muñoz-Rivas, Anabel Franco-Moreno, Eva Moya-Mateo, María Pilar Ferrera-Ahijado, Eva María Duro-Perales, María Antón-Vallejo, Montserrat Saiz-González, Alberto Diez-García, Celeste Marina-Verde, Andrés Rosa-López, Inmaculada Fernández-Sotillo, Carolina Espejo-Paeres, Miguel Bravo-Prieto, Javier Corrochano Del Pino, Carmen Pantoja-Zarza, María José Crespo-Carballés","doi":"10.1093/ehjdh/ztag129","DOIUrl":"https://doi.org/10.1093/ehjdh/ztag129","url":null,"abstract":"<p><strong>Aims: </strong>Current cardiovascular risk scores may underestimate the risk of future cardiovascular events, particularly in younger individuals with prolonged exposure to risk factors. In contrast, imaging-based detection of subclinical atherosclerosis provides a more accurate assessment of cardiovascular risk by identifying established vascular disease. We aimed to develop and prospectively validate an artificial intelligence-based tool using retinal fundus images to detect ultrasound-confirmed subclinical atherosclerosis.</p><p><strong>Methods and results: </strong>In this prospective observational study, 931 participants (mean age 52.6 years; 70.2% women) without prior cardiovascular disease underwent standardized clinical evaluation, non-mydriatic retinal imaging, and carotid and femoral ultrasound to detect subclinical atherosclerosis. A multimodal AI model integrating deep learning from retinal images with radiomic and clinical data was developed in a derivation cohort (<i>n</i> = 781) and evaluated in a held-out prospective test set (<i>n</i> = 150).Subclinical atherosclerosis was present in 50.8% of participants. In the prospective test set, the AI model demonstrated good discrimination. In image-only mode, the model achieved an area under the curve (AUC) of 0.80 (95% CI 0.73-0.87), with sensitivity of 88.2%. In the enhanced mode incorporating clinical variables, performance improved to an AUC of 0.86 (95% CI 0.80-0.92), with sensitivity of 93.4% and a negative predictive value of 90.6%. Discrimination was higher in younger individuals and those at low-to-intermediate cardiovascular risk.</p><p><strong>Conclusion: </strong>AI-based retinal image analysis enables non-invasive detection of systemic subclinical atherosclerosis. This scalable approach may enhance early identification of high cardiovascular-risk patients, particularly in populations in whom risk is underestimated.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag129"},"PeriodicalIF":4.4,"publicationDate":"2026-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13492337/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148802378","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Decentralized community-based hub-intermediary-spoke model for rapid cardiac ultrasound triage for early heart failure detection: findings from the Heart2Miss initiative.","authors":"Diana Hui Ping Foo, John Jui Ping Yeo, Yenny Yen Yen Yeo, Stephanie Tayas Bumphray, Rose Hui Chin Jong, Liana Lantong Sumbu, Claudia Lennya Jana, Jennett Michael, Maziah Ishak, Peter Jerampang, Maila Mustapha, Sally Suriani Ahip, Lenny Hamden, Macnicholson Igo, Mohammad Nor Azlan Sulaiman, Jawing Chunggat, Francesca Gelang Chong, Deborah Berenai Enggong, Yeequant Chung, Daniel Ishak, Angela Melinda Anthony Jalin, Sonesh Kalra, Alan Yean Yip Fong","doi":"10.1093/ehjdh/ztag115","DOIUrl":"10.1093/ehjdh/ztag115","url":null,"abstract":"<p><strong>Aims: </strong>To evaluate the feasibility and system-level performance of Heart2Miss, a decentralized community-based triage model deploying AI-powered point-of-care ultrasound (AI-POCUS) via a hub-intermediary-spoke approach in diabetes primary care for early heart-failure (HF) detection.</p><p><strong>Methods and results: </strong>In this prospective study, 1000 adults with diabetes and no known HF were screened over seven months across six primary care clinics (spokes); 985 with complete data were analysed. Novice biomedical and bioscience graduates underwent 4-week training to perform focused three-view handheld AI-POCUS. Images were AI-analysed and verified through the hub-intermediary-spoke pathway. The primary outcome was detection of previously undiagnosed HF. Secondary outcomes included reduction in tertiary-centre burden through the hub-intermediary-spoke pathway and novice sonographer performance. 11.1% (<i>n</i> = 109) had Stage B (pre-HF) and 1.0% (<i>n</i> = 10) Stage C HF (symptomatic HF). Rapid triage ruled out abnormality in 77.3% at the spoke and a further 12.6% after intermediary TTE confirmation, reducing tertiary diagnostic burden by 89.9%. Only 1.0% required tertiary referral. Regarding novice performance, >90% analysable scans were achieved for left-ventricular parameters and >85% for left-atrial volume. After 400 scans, scan time fell from 11.0 ± 5.3 min to 8.3 ± 4.4 min (Δ 2.31 min, 95% CI 1.52-3.11; <i>P</i> < 0.001), and complete three-view capture improved from 88.0% to 92.2% (<i>P</i> = 0.035).</p><p><strong>Conclusion: </strong>This decentralized hub-intermediary-spoke model combining AI-POCUS, telehealth verification, and a task-shifted bioscience workforce enabled early HF detection while substantially reducing specialist workload, supporting digital health-enabled workforce innovation and pathway redesign in resource-constrained settings.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag115"},"PeriodicalIF":4.4,"publicationDate":"2026-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13520925/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148842157","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-horizon machine learning for population-level hypertension risk stratification in the UK Biobank.","authors":"Akhil Naik, Ivan Olier, Ellen A Dawson, Garry McDowell, Deirdre A Lane, Gregory Y H Lip, Sandra Ortega-Martorell","doi":"10.1093/ehjdh/ztag128","DOIUrl":"10.1093/ehjdh/ztag128","url":null,"abstract":"<p><strong>Aims: </strong>Hypertension is a major contributor to cardiovascular morbidity and mortality, yet identifying individuals at risk before clinical diagnosis remains challenging. Here, we present a multi-horizon machine learning framework designed to model incident hypertension risk across multiple clinically meaningful time windows using data from 246 286 participants in the UK Biobank. The framework systematically compares predictive performance across five horizons under severe class imbalance, enabling analysis of how discrimination, precision, and risk drivers evolve as outcome prevalence changes over time.</p><p><strong>Methods and results: </strong>Seven classification algorithms were evaluated, including logistic regression, random forest, naïve Bayes, and four boosting-based ensemble methods. To enhance interpretability, we integrate SHapley Additive exPlanations (SHAP) with generative topographic mapping (GTM), combining feature-level attribution with population-level visualization of model predictions. Ensemble boosting models consistently achieved the strongest performance, with average precision increasing from 0.04 for the ≤2-year horizon to 0.22 for the ≤10-year horizon, while ROC-AUC remained relatively stable (∼0.75-0.79). Together, this framework reveals consistent predictors of hypertension risk (including baseline blood pressure, age, body mass index, medication burden, and cardiometabolic multimorbidity) and illustrates how combinations of risk factors organize hypertension risk across time horizons.</p><p><strong>Conclusions: </strong>Our results demonstrate how multi-horizon modelling and complementary explainability approaches can provide deeper insight into evolving disease risk patterns in large biomedical cohorts, supporting more interpretable and scalable strategies for population-level cardiovascular prevention. Such approaches may enable earlier identification of high-risk individuals and inform targeted screening and preventive interventions in routine care settings.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag128"},"PeriodicalIF":4.4,"publicationDate":"2026-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13505867/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148820458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-life clinical validation of artificial intelligence-assisted echocardiography for aortic root and left ventricular outflow tract measurements against human readers and cardiac magnetic resonance.","authors":"Patrycja Mołek-Dziadosz, Oleksandra Bondarchuk, Aleksandra Woźniak, Maria Królikowska, Agnieszka Mirek, Anna Furman-Niedziejko, Karol Mazur, Elżbieta Ostrowska-Kaim, Karolina Golińska-Grzybała, Aleksander Siniarski, Joanna Szachowicz-Jaworska, Tomasz Miszalski-Jamka, Radosław Rychlak, Marc R Dweck, Jadwiga Nessler, Andrzej Gackowski","doi":"10.1093/ehjdh/ztag126","DOIUrl":"10.1093/ehjdh/ztag126","url":null,"abstract":"<p><strong>Aims: </strong>Artificial intelligence (AI) algorithms may improve echocardiographic measurement standardization, but direct validation against cardiovascular magnetic resonance (CMR) remains limited. This study aimed to evaluate the agreement between AI-assisted transthoracic echocardiography (TTE), expert manual TTE, and CMR as a gold standard for aortic root and left ventricular outflow tract (LVOT) diameters.</p><p><strong>Methods and results: </strong>Out of 183 patients with analysable TTE and CMR recordings performed within 7 days, 169 had complete measurements of LVOT, sinotubular junction (STJ), and sinus of Valsalva (SoV) by AI and were included in the primary analysis. Automated AI measurements were obtained using the US2.ai platform and compared with those of expert echocardiographers and CMR. Agreement was assessed using intraclass correlation coefficients (ICC) and Bland-Altman analysis. Agreement between Expert 1 and CMR, and Expert 2 and CMR, yielded ICCs: 0.63 (95% CI: 0.50-0.73) and 0.51 (95% CI: 0.36-0.65) for LVOT, 0.71 (95% CI: 0.63-0.78) and 0.82 (95% CI: 0.73-0.87) for STJ, and 0.85 (95% CI: 0.79-0.89) to 0.85 (95% CI: 0.78-0.89) for SoV, respectively. Artificial intelligence-derived measurements demonstrated comparable agreement with CMR: ICC 0.70 (95% CI: 0.61-0.77) for LVOT, 0.77 (95% 0.70-0.83) for STJ, and 0.77 (95% 0.70-0.83) for SoV. Clinically significant differences (≥2 mm for LVOT and ≥4 mm for STJ and SoV) between AI and CMR measurements were observed in 36.7% of LVOT, 20% of STJ, and 11.8% of SoV measurements.</p><p><strong>Conclusion: </strong>Artificial intelligence-assisted echocardiography showed comparable agreement for aortic root and LVOT measurements, but physician oversight remains necessary.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag126"},"PeriodicalIF":4.4,"publicationDate":"2026-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13505863/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148820498","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable electrocardiogram interpretation using deep learning-based semantic segmentation.","authors":"Bauke K O Arends, Bas B S Schots, Parmenion Koutsogeorgos, Timo Nijkamp, Tim M Paquaij, Diantha J M Schipaanboord, Rutger J Hassink, Pim van der Harst, René van Es, Rutger R van de Leur","doi":"10.1093/ehjdh/ztag122","DOIUrl":"10.1093/ehjdh/ztag122","url":null,"abstract":"<p><strong>Aims: </strong>Accurate electrocardiogram (ECG) waveform delineation, rhythm classification, and median beat generation are interdependent steps whose joint modelling improves consistency for downstream computerized diagnostic tasks. This study aimed to develop a lead-agnostic segmentation model that performs these tasks by segmenting individual leads and aggregating predictions in post-processing.</p><p><strong>Methods and results: </strong>A DeepLabV3-based neural network was trained to segment ECG leads into 20 waveform and rhythm classes using 1931 annotated ECGs and 33 093 ECGs with physician-verified diagnostic statements. Post-processing combined lead-wise predictions to delineate intervals, classify rhythm, and construct median beats. Performance was evaluated on internal (<i>n</i> = 988) and external (<i>n</i> = 1303) test sets. Median beats were compared with PTB-XL+ reference medians from Marquette 12SL and University of Glasgow (Uni-G) using similarity metrics and downstream classification. An interactive web tool (http://segmentation.ecgx.ai) was released to support further research. In the external set, delineation for PQ interval, QRS duration, and QT interval had mean errors of -0.9 ± 10.4 ms, 0.2 ± 7.4 ms, and 1.8 ± 16.0 ms. Rhythm classification was performed by assigning class labels to segmented waveform components, with weighted F1 scores of 0.94 for P waves and 0.89 for QRS complexes. Qualitative review showed good alignment with reference beats and differences in beat selection, adjacent beat handling, and QRS identification. Downstream classification performance was equivalent to Marquette 12SL medians but statistically superior to Uni-G medians for 7/10 diagnostic labels.</p><p><strong>Conclusion: </strong>This study demonstrates a robust, clinically applicable, vendor- and lead-agnostic deep learning model for ECG analysis, encompassing waveform delineation, rhythm classification, and median beat construction.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag122"},"PeriodicalIF":4.4,"publicationDate":"2026-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528239/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of an LLM pipeline exceeding physician-documented cardiovascular risk scores under routine clinical conditions.","authors":"Tobias Roeschl, Marie Hoffmann, Axel Unbehaun, Henryk Dreger, Gerhard Hindricks, Volkmar Falk, Ran Balicer, Radu Tanacli, Felix Hohendanner, Alexander Meyer","doi":"10.1093/ehjdh/ztag124","DOIUrl":"https://doi.org/10.1093/ehjdh/ztag124","url":null,"abstract":"<p><strong>Aims: </strong>Risk scores are essential to evidence-based cardiovascular care, but manual calculation is labour intensive and error prone. Large language models (LLMs) could automate this process, yet LLMs are limited by their propensity for calculation errors and factual hallucinations. Pipelines separating LLM-based data extraction from deterministic score computation may improve reliability and transparency.</p><p><strong>Methods and results: </strong>We conducted a retrospective diagnostic study at a quaternary heart centre in Germany (January 2020 to July 2023). Patients with atrial fibrillation (<i>n</i> = 179) from an ablation registry and patients with severe aortic stenosis (<i>n</i> = 76) evaluated by a heart team were included. Six LLMs (GPT-5.2, Gemini 3.1 Pro, DeepSeek-R1, Qwen3, GPT-OSS 120B, and Kimi K2.5) were tested in standalone, retrieval-augmented generation (RAG), and pipeline configurations to compute HAS-BLED, CHA₂DS₂-VASc, and EuroSCORE II scores from routine clinical reports. Accuracy was assessed against expert-adjudicated ground truth using root mean squared error (RMSE) and Krippendorff's α to evaluate numerical deviation and categorical agreement, respectively. Pipeline-generated scores showed substantially higher agreement with expert adjudication than standalone LLMs, LLMs with RAG, and treating physicians (mean Krippendorff's α: 0.78 vs. 0.32 vs. 0.39 vs. 0.31) and lower deviation from ground truth (mean RMSE: 0.89 vs. 5.81 vs. 1.85 vs. 1.34).</p><p><strong>Conclusion: </strong>Pipelines combining expert-curated knowledge injection, LLM-based clinical data extraction, and deterministic score calculation enable accurate and scalable cardiovascular risk score computation from unstructured real-world clinical data, outperforming physician-documented scores. Such pipelines could form the basis for clinical decision-support systems that automate routine risk assessment, reduce clinician workload, and promote more consistent evidence-based care.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag124"},"PeriodicalIF":4.4,"publicationDate":"2026-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13492338/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148802422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}