European heart journal. Digital health最新文献

{"title":"Digital biomarkers in cardiovascular medicine: a scientific statement of the ESC Working Group on e-Cardiology, the ESC Working Group on Cardiovascular Pharmacotherapy, the European Heart Rhythm Association (EHRA) of the ESC, the Association for Acute Cardiovascular Care (ACVC) of the ESC, the Heart Failure Association (HFA) of the ESC, the European Association of Preventive Cardiology (EAPC) of the ESC, the European Association of Cardiovascular Imaging (EACVI) of the ESC, in collaboration with the Digital Cardiology and Artificial Intelligence Committee of the ESC, and the ESC Regulatory Affairs Committee.","authors":"Pablo M Corredoira, Juan Carlos Kaski, David Duncker, Marcus Dörr, Matthias Wilhelm, Marco Tubaro, Mark J Schuuring, Mamas Mamas, Steffen E Petersen, Rafael Vidal-Pérez, Hareld M C Kemps, Emanuela T Locati, Christian Mueller, Magnus T Jensen, Raffaele De Lucia, Loreena Hill, Julia Ramírez, Matthijs Cluitmans, Nurgül Keser, Axel Verstrael, Polychronis E Dilaveris, Paul A Friedman, Rubén Casado-Arroyo","doi":"10.1093/ehjdh/ztag074","DOIUrl":"10.1093/ehjdh/ztag074","url":null,"abstract":"<p><p>Cardiovascular disease is the leading cause of morbidity and mortality globally, imposing a substantial patient burden. Digital biomarkers derived from digital health technologies enable quantifiable, near-continuous, real-world capture of pathophysiological signals and are increasingly used across cardiovascular medicine. Their growth is often supported by big data analytics, artificial intelligence and connected sensors within the Internet of Things. This scientific statement systematically reviews digital biomarkers, their clinical utility across the cardiovascular care continuum, and summarizes major methodological and implementation challenges. We searched PubMed/MEDLINE, Embase and Scopus for phase III-IV cardiovascular trials published between 1 January 2019 and 1 August 2024 that collected digital biomarkers. We identified 541 records and included 32 trials (40 reports) enrolling 32 246 participants (44.71% women; mean age 66.2 ± 11.0 years). Heart failure was the most frequent target condition (16 trials, 50%), followed by cardiac implantable electronic devices trials (12 trials, 38%). Digital biomarkers included cardiac rhythm and heart rate, blood pressure, body weight, impedance-based measures, haemodynamic pressures, and physical activity and sleep patterns captured using diverse sensor technologies. Methodological quality was good (modal quality score 5; range 3-7), although most trials were unblinded. Current studies suggest potential clinical utility in selected cardiovascular conditions. Implementation remains challenging because of measurement variability, privacy and regulatory requirements, affordability and reimbursement barriers and inequities in access. Robust validation and outcome- and cost-effectiveness evidence are still needed. Registration number: CRD42024573879.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag074"},"PeriodicalIF":4.4,"publicationDate":"2026-07-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13378772/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148581029","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":"Foundation models for cardiovascular diseases and medicine.","authors":"Jiancheng Ye, Sophie Bronstein, Malak Abu Hashish","doi":"10.1093/ehjdh/ztag113","DOIUrl":"10.1093/ehjdh/ztag113","url":null,"abstract":"<p><p>Cardiovascular disease (CVD) remains the leading cause of mortality worldwide, accounting for more than 17 million deaths annually. Foundation models (FMs) are large-scale artificial intelligence systems pre-trained on broad datasets and increasingly studied for cardiovascular diagnosis, risk prediction, workflow support, and treatment personalization. This narrative review synthesizes evidence on FMs applied to cardiovascular medicine, including text-based, image-based, multimodal, and time-series models. The review examined model architectures, training strategies, performance metrics, clinical applications, implementation challenges, and evidence maturity using a three-tier framework. Cardiovascular FMs show promising but unevenly validated capabilities across multiple applications. Electrocardiogram (ECG)-based models such as ECGFounder have reported strong external validation performance for rhythm and conduction diagnoses, whereas imaging models such as Echo-Vision-FM support cardiac function assessment and pathology classification. Multimodal models, including CardioGPT and EchoCLIP, illustrate the potential value of integrating clinical text, imaging, and physiological signals, although several remain at a conceptual or proof-of-concept stage. Foundation models may become important tools in cardiovascular medicine by supporting earlier diagnosis, risk stratification, and clinical decision-making. However, the current evidence base remains heterogeneous, and major barriers persist, including limited interpretability, underrepresentation of diverse populations in training data, regulatory uncertainty, computational demands, and workflow integration challenges. Inclusive data development, transparent reporting, external validation, prospective clinical evaluation, and interdisciplinary implementation research are needed before these models can be deployed safely and equitably in routine cardiovascular care.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag113"},"PeriodicalIF":4.4,"publicationDate":"2026-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528242/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868196","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":"Role of artificial intelligence in developing predictive models for major adverse cardiovascular outcomes using CCTA adipose tissue characteristics: a systematic review and meta-analysis.","authors":"Sadaf Salehi, Seyed Hesam Hojjat, Ali Samadi Shams, Sepehr Ramezanipour, Armina Farkarian, Ali Azizi, Ghazaleh Kokabi Ghahremanpour, Hossein Movahed, Nikta Heidari, Soha Amiri, Aref Ghanaatpisheh","doi":"10.1093/ehjdh/ztag104","DOIUrl":"10.1093/ehjdh/ztag104","url":null,"abstract":"<p><p>Cardiovascular diseases remain as a leading cause of mortality and morbidity worldwide, with coronary artery disease (CAD) and its complications, collectively referred to as major adverse cardiovascular events (MACEs), necessitating accurate risk stratification. Coronary computed tomography angiography (CCTA) has emerged as a valuable non-invasive imaging modality, and adipose tissue characteristics derived from CCTA have shown promise as imaging biomarkers for MACE prediction. This systematic review and meta-analysis aimed to evaluate the predictive performance of artificial intelligence (AI)-driven models incorporating CCTA-derived adipose tissue radiomic features for forecasting MACEs. A systematic review and random-effects meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Studies evaluating predictive models developed using different AI algorithms that utilized adipose tissue radiomics as the primary predictor of MACEs in patients undergoing CCTA were included. Model performance was assessed using pooled area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Eleven studies comprising 47 244 participants were included in the analysis. AI-based models integrating adipose tissue radiomic features with clinical data consistently outperformed conventional risk assessment tools, with pooled AUCs ranging from 82.2% to 87.9%. Among the evaluated approaches, deep learning models demonstrated superior predictive performance compared with traditional machine learning and logistic regression-based models. However, considerable heterogeneity (<i>I</i>² > 96%) was observed across studies, reflecting variations in study design, imaging protocols, and AI methodologies. While AI-enhanced CCTA-based adipose tissue characterization demonstrates considerable potential for improving MACE risk prediction, methodological heterogeneity and limited external validation currently restrict its clinical applicability. Future research should prioritize standardized imaging and analytical methodologies, rigorous clinical and external validation, and transparent reporting to facilitate reliable integration of these AI models into routine cardiovascular risk assessment.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag104"},"PeriodicalIF":4.4,"publicationDate":"2026-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13367595/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148450843","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":"The current state of artificial intelligence-based invasive coronary angiography image analysis: a systematic review.","authors":"Viktor O van der Valk, Wouter P M van der Loo, Nico Bruining, Jouke Dijkstra, Douwe E Atsma, Marius Staring, Roderick W C Scherptong","doi":"10.1093/ehjdh/ztag110","DOIUrl":"10.1093/ehjdh/ztag110","url":null,"abstract":"<p><p>Invasive coronary angiography (ICA) is the reference standard for diagnosing coronary artery disease and guiding percutaneous coronary intervention, yet clinical interpretation relies largely on visual assessment, which is variable and often requires additional invasive testing to assess functional significance. Artificial intelligence (AI)-based analysis of ICA images has emerged as a potential solution to automate interpretation, improve reproducibility, and extract anatomical and physiological information directly from angiograms. We conducted a systematic review of AI applications for ICA image analysis, registered in PROSPERO and reported according to PRISMA guidelines. A total of 134 studies were included, covering tasks across the ICA workflow, including automated frame selection, vessel segmentation, lesion detection and quantification, prediction of invasive physiological indices, coronary anatomy labelling, image registration and reconstruction, outcome prediction and left ventricular function estimation. Most studies focused on vessel segmentation and lesion assessment, generally demonstrating high internal performance but marked heterogeneity in datasets, reference standards, evaluation metrics, and validation strategies. While earlier work relied predominantly on single-centre retrospective validation, more recent studies increasingly incorporate multi-centre data, external validation and prospective evaluation. AI-based prediction of invasive physiological indices appears particularly promising for reducing reliance on wire-based measurements, though robust prospective evaluation remains limited. Overall, AI-based ICA analysis has progressed from technical feasibility studies towards clinically oriented applications. However, challenges in generalizability, methodological standardization, and workflow integration must be addressed to enable reliable clinical adoption.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag110"},"PeriodicalIF":4.4,"publicationDate":"2026-07-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13387071/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148550920","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":"Systematic reviews of medical machine learning: limitations of pooling AUCs.","authors":"Amulya Gupta, Rahel Sileshi, Jeffrey Thompson, Amit Noheria","doi":"10.1093/ehjdh/ztag107","DOIUrl":"10.1093/ehjdh/ztag107","url":null,"abstract":"<p><p>The medical literature has seen rapid growth in studies that develop and evaluate machine learning (ML) classifiers for diagnostics, accompanied by an increase in systematic reviews for synthesizing this evidence. Many such reviews use meta-analyses that pool performance metrics such as area under the receiver-operating characteristic curve (AUC), sensitivity, and specificity across ML studies. In our view, quantitative pooling of such metrics is statistically and conceptually inappropriate in most settings, except when the same model is being tested across different samples. ML classifiers differ fundamentally in training data, model specification, and validation strategies. As a result, there is no underlying 'true AUC' that pooling aims to recover. In addition, AUC is a sampling-dependent, non-linear, and bounded measure whose value depends on relative ranking within a dataset, which makes it unsuitable for aggregation. Pooling sensitivity and specificity presents similar problems due to threshold dependence, inconsistent reporting practices, and inappropriate weighting by test-set size. Outside limited scenarios in which aggregation may be reasonable, undue emphasis on pooled performance estimates obfuscates critical descriptors of model validity and generalizability. We propose that systematic reviews of medical ML models prioritize structured, descriptive synthesis of key study characteristics, including dataset composition, validation strategy, input modalities, outcome definitions, data augmentation, independent validation, and deployability. When summary performance measures are presented, they are better presented as stratified summaries in subgroups of comparable studies. A concerted discourse on standardizing reporting standards is essential to guide systematic reviews that meaningfully inform the clinical utility of ML models.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag107"},"PeriodicalIF":4.4,"publicationDate":"2026-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13372031/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148457619","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 model for early risk stratification of 28-day mortality after myocardial infarction.","authors":"Pierre G Aublin, Stephanie G Kühne, David Füller, Benjamin Sasko, Hannah Schulze, Oliver Ritter, Ferdinand Bauke, Christine Meisinger, Jakob Linseisen, Philip Raake, Sebastian Zaunseder, Timo Schmitz, Dario Bongiovanni","doi":"10.1093/ehjdh/ztag079","DOIUrl":"10.1093/ehjdh/ztag079","url":null,"abstract":"<p><strong>Aims: </strong>Despite significant improvement in transcatheter therapies, myocardial infarction (MI) retains a high burden of morbidity and mortality. While several scores have been developed, an established risk model tailored to predict short-term mortality after MI is lacking.</p><p><strong>Methods and results: </strong>In this study, we leveraged real-world data to develop a machine learning (ML) model for predicting 28-day mortality after MI, relying on clinical data collected within the first 24 h of hospitalization. From the Augsburg Myocardial Infarction Registry, 14 725 MI patients surviving the first 24 h of the acute event were included: 80% in the training set for model development to train a Gradient Boosting Machine while the 20% remaining served as test set. A sequential feature selection process retained variables as long as their addition improved AUPRC within 5-fold cross-validation on the train set. We then assessed our model on the test set and performed an external validation on 1328 patients from the Brandenburg Myocardial Infarction Registry. Shapley additive explanations values were used to explain the ML model's predictions. Our model retained six variables including age, systolic blood pressure, admission levels of glucose, creatinine, leucocytes, and whether the patient had a prehospital cardiac arrest. It exhibited good calibration and discrimination ability on the internal (AUROC=0.82) and external validation (AUROC=0.85) and outperformed common scores like GRACE, CADILLAC, and TIMI (<i>P</i> = 0.05).</p><p><strong>Conclusion: </strong>The results highlight the potential of our model to ease the detection of patients at high risk, allowing targeted early intervention to improve their outcome, and to contribute to improved precision medicine.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag079"},"PeriodicalIF":4.4,"publicationDate":"2026-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13335799/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148439122","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":"Cardiology hospital admission risk prediction: training, internal validation and technical implementation in the electronic health record.","authors":"Jasper L Selder, Olivier V Witteman, Oscar M van der Meer, Machteld J Boonstra, Stephan van der Zwaard, Folkert W Asselbergs, Marlies P Schijven, Albert C van Rossum, Cornelis P Allaart","doi":"10.1093/ehjdh/ztag109","DOIUrl":"10.1093/ehjdh/ztag109","url":null,"abstract":"<p><strong>Aims: </strong>Rising healthcare demand is increasingly outpacing available outpatient capacity in cardiology, where follow-up is often scheduled at fixed intervals despite substantial variation in individual patient risk. This uniform follow-up approach contributes to high outpatient workload and inefficient use of clinical resources. Accurate risk estimation using routinely collected electronic health record (EHR) data may support more individualized follow-up planning by identifying patients at very low risk of mortality or unplanned hospitalization, in whom follow-up intervals could be safely extended.</p><p><strong>Methods and results: </strong>We developed and validated a machine-learning model as part of the Cardiology Hospital Admission Risk Prediction (CHARP) program. The retrospective baseline cohort comprised 307 792 outpatient visits from 52 989 unique patients at Amsterdam UMC. The primary endpoint was a composite of unplanned cardiac hospitalization or all-cause death within 2 years; the 1-year composite endpoint served as a secondary outcome. Model development and validation were performed in a filtered, leakage-safe prediction cohort using gradient-boosted decision trees (XGBoost) with strict patient-level GroupKFold cross-validation. All predictions and performance metrics were retrospectively evaluated at the outpatient visit (trigger) level. Retrospective model performance was assessed using AUROC, AUPRC, Brier score, calibration curves, and SHAP-based explainability. The final model was technically deployed within the electronic health record to allow automated, visit-level risk estimation in a prospective silent-running environment. In the filtered prediction cohort (199 961 visits), the 2-year composite endpoint prevalence was 16.8%. Across five cross-validation folds, the CHARP model achieved a mean AUROC of 0.77 ± 0.00 and AUPRC of 0.42 ± 0.01, with a Brier score of 0.12, indicating strong overall discrimination and good calibration. Key predictors included NT-proBNP, renal function indices, prior hospitalizations, and cardiac function measures. The deployed CHARP pipeline successfully generated daily risk predictions for all scheduled cardiology outpatients in the EHR environment throughout the silent-running period.</p><p><strong>Conclusion: </strong>This study shows that machine-learning applied to routine EHR data can deliver clinically meaningful, visit-level risk stratification for cardiology outpatients. The successful EHR integration of CHARP enables prospective evaluation of data-driven follow-up strategies aimed at reducing outpatient clinic burden through safe de-intensification of follow-up for low-risk patients.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 7","pages":"ztag109"},"PeriodicalIF":4.4,"publicationDate":"2026-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13419068/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148622894","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":"Extended reality simulator for dynamic visualization and evaluation of stereotactic arrhythmia radioablation (STAR) treatment plans within RAVENTA trial.","authors":"Domenico Riggio, Joana Leitão, Melanie Grehn, Judit Boda-Heggemann, Adrian Zaman, João Seco, Oliver Blanck, Maria Francesca Spadea","doi":"10.1093/ehjdh/ztag100","DOIUrl":"10.1093/ehjdh/ztag100","url":null,"abstract":"<p><p>Stereotactic arrhythmia radioablation (STAR) represents an emerging non-invasive treatment for therapy-refractory ventricular tachycardia. Yet, planning remains challenged by multimodal cardiac imaging integration, electroanatomical mapping (EAM) transfer, and cardiorespiratory motion effects on dose delivery. Current radiotherapy (RT) planning systems offer mainly static visualization and limited access to intramural myocardial structures, hindering communication between cardiology and radiation oncology teams. We present a novel extended reality (XR) simulator designed to dynamically visualize STAR-relevant imaging and planning data. The system integrates diastolic cardiac CT, respiratory-binned 4DCT, anatomical segmentations, EAM data, and phase-recomputed RT dose distributions within an XR environment. Cardiac structures are propagated across respiratory phases using deformable registration, while dose distributions are recomputed on each respiratory-binned CT, enabling phase-specific inspection of dose conformality for both planning target volumes (PTVs) and cardiac target volumes (CardTVs). The resulting time-resolved volumetric dataset is rendered in XR, allowing clinicians to explore cardiac motion, visualize intramural dose deposition, and jointly assess target and organ-at-risk dynamics. This supports qualitative evaluation of dose-motion interplay and interdisciplinary interpretation of intramural targets. The system was tested on three STAR patients enrolled in the RAVENTA trial. Motion analysis revealed PTV centroid displacement amplitudes over the breathing cycle of up to 17.5, 10.2, and 8.9 mm for patients 1, 2, and 3, respectively, with conformity number variations of 0.38, 0.34, and 0.19. Expert evaluation showed positive perceived utility for target-anatomy-dose understanding, motion interpretation, and multidisciplinary communication. This proof-of-concept demonstrates the feasibility and potential clinical value of XR-based motion-aware dose visualization for STAR planning.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag100"},"PeriodicalIF":4.4,"publicationDate":"2026-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13387070/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148551725","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":"Electronic health records research: easy evidence, harder oversight.","authors":"Abdalhakim Shubietah, Kerollos Abdelsayed, Mohammed Ruzieh","doi":"10.1093/ehjdh/ztag105","DOIUrl":"10.1093/ehjdh/ztag105","url":null,"abstract":"<p><p>Electronic health record (EHR) research networks have accelerated clinical research by making it possible to study large and diverse patient populations over time. However, the quality and resources required for these studies vary widely, depending on whether they rely on simple aggregate outputs or a more rigorous patient-level data, a central but occasionally unstated distinction in EHR-based research. Patient-level data can support a more reproducible cohort construction, time-anchored exposure definitions, and more precise outcome ascertainment, whereas aggregate approaches limit rigour regarding the precise definition of index dates, the completeness of covariate capture, the accurate characterization of medication exposure windows, and the validity of outcome ascertainment. These limitations may amplify well-known threats to causal inference in observational research, including misclassification, incomplete outcome capture across health systems, residual confounding, and immortal time bias, particularly when comparative effectiveness questions are pursued. In this viewpoint, we outline why 'easy evidence' can yield fragile conclusions when analytic granularity is limited, and we propose pragmatic expectations for reporting and review: explicit declaration of data level (patient vs. aggregate), clear time-zero definitions for all arms, validated case-definition algorithms when available, sensitivity analyses aligned with plausible misclassification, and transparency sufficient to enable reproducibility. Strengthening methodological norms is essential to preserve the credibility of EHR-based evidence as its influence on clinical discourse continues to grow.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag105"},"PeriodicalIF":4.4,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13353525/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148426994","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":"Rational and design of the digital diagnosis of cardiac sounds in paediatric patients (DI_SOUND) study.","authors":"Gabriele Egidy Assenza, Alessandro Colombo, Vittoria Mastromarino, Angela D'Aversa, Ylenia Bartolacelli, Simone Bonetti, Giulio Calcagni, Luigi Corvaglia, Luca Di Ienno, Andrea Donti, Maria Francesca Fierro, Matteo Matteucci, Raffaella Marzullo, Carolina Putotto, Maria Giovanna Russo, Giuliana Simonazzi, Sara Bego, Paolo Versacci, Bruno Marino Taussig de Bodonia","doi":"10.1093/ehjdh/ztag103","DOIUrl":"10.1093/ehjdh/ztag103","url":null,"abstract":"<p><strong>Aims: </strong>Congenital heart diseases (CHD) are the most common birth defects in humans. Current clinical newborn screening for CHD has limitations. Digital elaboration of cardiac sounds with diagnostic purposes has been explored in the recent past in adults and older children. We propose to develop a dedicated software for automatic dichotomous clinical classification of heart sounds in newborns.</p><p><strong>Methods and results: </strong>The Digital Diagnosis of cardiac SOUNDs in paediatric patients (DI_SOUND) study is a multicentre study with two sequential phases: (i) derivation/training and (ii) validation. Each newborn will undergo a digital recording of cardiac sounds (stored as wav file using the Littmann Core Stethoscope) and comprehensive echocardiogram. In the development phase, sound tracings will undergo processing to remove background noise. They will be segmented to generate sound samples of homogeneous suitable tuned duration, and filtered records will be normalized for preventing inhomogeneity in the extracted features. Data analysis will be then implemented based on handcrafted feature extraction. A DI_SOUND classifier algorithm will be developed (<b>Aim</b> 1). In the validation phase, consecutive newborns will undergo clinical (standard) screening and digital screening using the DI_SOUND algorithm. Comprehensive echocardiogram will serve as the reference to assess diagnostic performance of the DI_SOUND algorithm (Aim 2). A cost-effective analysis will be used to model marginal cost-effectiveness of digital screening using clinical screening as comparison (Aim 3).</p><p><strong>Conclusion: </strong>The study will prove if development and validation of a digital classifier of neonatal cardiac sounds is feasible and how this will impact clinical outcomes, patient care, hospital admissions, and screening efficiency.</p><p><strong>Clinicaltrialsgov id: </strong>NCT07542509.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":"7 6","pages":"ztag103"},"PeriodicalIF":4.4,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13361984/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148439257","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}