@Article{info:doi/10.2196/76215, author="Yu, Min-Young and Yoo, Young Hae and Han, In Ga and Kim, Eun-Jung and Son, Youn-Jung", title="Comparing the Performance of Machine Learning Models and Conventional Risk Scores for Predicting Major Adverse Cardiovascular Cerebrovascular Events After Percutaneous Coronary Intervention in Patients With Acute Myocardial Infarction: Systematic Review and Meta-Analysis", journal="J Med Internet Res", year="2025", month="Jul", day="18", volume="27", pages="e76215", keywords="machine learning", keywords="mortality", keywords="myocardial infarction", keywords="patient readmission", keywords="percutaneous coronary intervention", keywords="prediction algorithm", keywords="statistical model", abstract="Background: Machine learning (ML) models may offer greater clinical utility than conventional risk scores, such as the Thrombolysis in Myocardial Infarction (TIMI) and Global Registry of Acute Coronary Events (GRACE) risk scores. However, there is a lack of knowledge on whether ML or traditional models are better at predicting the risk of major adverse cardiovascular and cerebrovascular events (MACCEs) in patients with acute myocardial infarction (AMI) who have undergone percutaneous coronary interventions (PCI). Objective: The aim of this study is to systematically review and critically appraise studies comparing the performance of ML models and conventional risk scores for predicting MACCEs in patients with AMI who have undergone PCI. Methods: Nine academic and electronic databases including PubMed, CINAHL, Embase, Web of Science, Scopus, ACM, IEEE, Cochrane, and Google Scholar were systematically searched from January 1, 2010, to December 31, 2024. We included studies of patients with AMI who underwent PCI, and predicted MACCE risk using ML algorithms or conventional risk scores. We excluded conference abstracts, gray literature, reviews, case reports, editorials, qualitative studies, secondary data analyses, and non-English publications. Our systematic search yielded 10 retrospective studies, with a total sample size of 89,702 individuals. Three validation tools were used to assess the validity of the published prediction models. Most included studies were assessed as having a low overall risk of bias. Results: The most frequently used ML algorithms were random forest (n=8) and logistic regression (n=6), while the most used conventional risk scores were GRACE (n=8) and TIMI (n=4). The most common MACCEs component was 1-year mortality (n=3), followed by 30-day mortality (n=2) and in-hospital mortality (n=2). Our meta-analysis demonstrated that ML-based models (area under the receiver operating characteristic curve: 0.88, 95\% CI 0.86?0.90; I{\texttwosuperior}=97.8\%; P<.001) outperformed conventional risk scores (area under the receiver operating characteristic curve: 0.79, 95\% CI 0.75?0.84; I{\texttwosuperior}=99.6\%; P<.001) in predicting mortality risk among patients with AMI who underwent PCI. Heterogeneity across studies was high. Publication bias was assessed using a funnel plot. The top-ranked predictors of mortality in both ML and conventional risk scores were age, systolic blood pressure, and Killip class. Conclusions: This review demonstrated that ML-based models had superior discriminatory performance compared to conventional risk scores for predicting MACCEs in patients with AMI who had undergone PCI. The most commonly used predictors were confined to nonmodifiable clinical characteristics. Therefore, health care professionals should understand the advantages and limitations of ML algorithms and conventional risk scores before applying them in clinical practice. We highlight the importance of incorporating modifiable factors---including psychosocial and behavioral variables---into prediction models for MACCEs following PCI in patients with AMI. In addition, further multicenter prospective studies with external validation are required to address validation limitations. Trial Registration: PROSPERO CRD42024557418; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024557418 ", doi="10.2196/76215", url="https://www.jmir.org/2025/1/e76215" } @Article{info:doi/10.2196/71675, author="Olatoye, Isaac Toba", title="Discovery of Novel Inhibitors of HMG-CoA Reductase Using Bioactive Compounds Isolated From Cochlospermum Species Through Computational Methods: Virtual Screening and Algorithm Validation Study", journal="JMIRx Bio", year="2025", month="Jul", day="10", volume="3", pages="e71675", keywords="HMGR", keywords="statins", keywords="hypercholesterolemia", keywords="cochlospermum", keywords="phytochemicals", keywords="molecular docking", keywords="3-hydroxy-3-methylglutaryl coenzyme-A reductase", abstract="Background: Cholesterol biosynthesis is a critical pathway in cellular metabolism, with 3-hydroxy-3-methylglutaryl coenzyme-A reductase (HMGR) catalyzing its committed step. HMGR inhibition has been widely explored as a therapeutic target for managing hypercholesterolemia, and statins are the most commonly used competitive inhibitors. However, the search for novel, natural HMGR inhibitors remains a vital area of research, due to the adverse effects associated with long-term statin use. Cochlospermum planchonii and Cochlospermum tinctorium are West African medicinal plants traditionally used to treat metabolic disorders, including dyslipidemia. Despite their usefulness, the specific bioactive compounds responsible for these effects are currently poorly characterized, justifying further investigations. Objective: This study investigates the potential of phytochemicals from Cochlospermum planchonii and Cochlospermum tinctorium as natural inhibitors of human HMGR using molecular docking techniques. Methods: A total of 84 phytochemicals from 2 species of Cochlospermum as reported in literature, were evaluated as potential inhibitors of HMGR. Using DataWarrior software, their drug-likeness and ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties were screened in accordance with Lipinski's Rule of Five. The 32 compounds that met the criteria were docked on PyRx against the HMG-binding site of HMGR, alongside atorvastatin (native ligand) and 6 known statins, which served as control ligands. Results: Docking analysis of their two best binding modes showed that 10 (31.3\%) out of the 32 screened phytochemicals demonstrated strong binding affinities and interactions with the HMG-binding pocket (residues 682?694) of HMGR, with binding energy ($\Delta$G) scores ranging from ?4.6 to ?6.0 kcal/mol, comparable to or exceeding those of statins (?4.6 to ?5.7 kcal/mol). Their docking scores (?13.272 to ?32.103) also compared favorably with those of statins (?25.939 to ?36.584). Interestingly, 3-O-methylellagic acid (ID\_13915428) demonstrated the strongest interaction, forming 26 binding interactions with the HMG-binding pocket residues, more than any compound, including statins. One-way ANOVA of the mean and SEM of the binding affinity scores for the phytochemicals and statins (9 replicates each) indicated a statistically significant difference at P<.05 (total sample size n=153; actual P=.0001). Conclusions: This study is the first to virtually screen and identify specific bioactive compounds isolated from Cochlospermum planchonii and Cochlospermum tinctorium with potential cholesterol-lowering effects in humans. The findings not only support the traditional use of these plants in West Africa to manage dyslipidemia and other ailments, but also present the phytochemicals as promising drug candidates for further optimization as natural inhibitors of HMGR. However, while this study provides valuable computational insights into the molecular interactions of the compounds with HMGR, further advanced computational, in vitro, and in vivo studies are still necessary to validate their inhibitory potential and therapeutic applications. ", doi="10.2196/71675", url="https://bio.jmirx.org/2025/1/e71675" } @Article{info:doi/10.2196/73389, author="Kim, Myung-Rho and Shaikh, Taha and Wang, Shawn and Taylor, Spencer and Goel, Vidhani and Khetarpal, Kaur Banveet and Ahsan, Chowdhury and Batra, Kavita", title="A Higher-Than-Standard-Intensity International Normalized Ratio Goal for Patients Undergoing Mechanical Aortic Valve Replacement With Additional Thrombotic Risk Factors: Protocol for a Systematic Review and Meta-Analysis", journal="JMIR Res Protoc", year="2025", month="Jul", day="10", volume="14", pages="e73389", keywords="mechanical aortic valve replacement", keywords="MAVR", keywords="thromboembolic risk factors", keywords="thromboembolism", keywords="anticoagulation", keywords="international normalized ratio", keywords="warfarin", keywords="Coumadin", keywords="vitamin K antagonist", abstract="Background: Lifelong anticoagulation therapy with vitamin K antagonists is recommended following mechanical aortic valve replacement (MAVR) to prevent valve thrombosis. Current guidelines recommend a standard international normalized ratio (INR) of 2.5 for patients with MAVR without additional thromboembolic risk factors, and a higher INR goal of 3.0 for those with conditions such as atrial fibrillation, prior thromboembolism, or left ventricular dysfunction. However, limited clinical evidence exists to guide anticoagulation intensity in this high-risk subgroup, necessitating a systematic review. Objective: We aimed to assess the safety and efficacy of higher-intensity INR goals (>3.0) compared to standard-intensity goals (approximately 2.5) in patients with MAVR with additional thromboembolic risk factors. Methods: This protocol describes a systematic review and meta-analysis following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. A comprehensive literature search will be conducted across PubMed, Embase, and the Cochrane Library for studies published before December 18, 2024. Eligible studies include randomized controlled trials (RCTs), cohort studies, and follow-up studies involving adult patients with MAVR on warfarin therapy, stratified by the presence of additional thromboembolic risk factors. Non--English-language studies, case reports, editorials, and animal studies will be excluded. Results: The review will synthesize existing data to compare the risks and benefits of intensified anticoagulation in patients with MAVR with additional thromboembolic risk factors. Data analysis and manuscript preparation are scheduled for July-August 2025. Conclusions: This study will provide critical evidence on INR management in high-risk patients with MAVR, potentially informing future updates to clinical guidelines and optimizing the balance between thromboembolic prevention and bleeding risk. International Registered Report Identifier (IRRID): PRR1-10.2196/73389 ", doi="10.2196/73389", url="https://www.researchprotocols.org/2025/1/e73389" } @Article{info:doi/10.2196/71062, author="Marrison, Tucker Sarah and Shungu, Nicholas and Diaz, Vanessa", title="Perception and Counseling for Cardiac Health in Breast Cancer Survivors Using the Health Belief Model: Qualitative Analysis", journal="JMIR Cancer", year="2025", month="Jul", day="3", volume="11", pages="e71062", keywords="cardiovascular health", keywords="cancer survivorship", keywords="lifestyle counseling", keywords="breast cancer", keywords="cancer survivors", abstract="Background: Breast cancer survivors have increased cardiovascular risk compared to those without cancer history. Cardiovascular disease is the most common cause of death in breast cancer survivors. Cardiovascular risk in breast cancer survivors is impacted by both cancer treatment--associated effects and in risk factors for breast cancer and cardiovascular disease overlap. Strategies to improve screening for and management of cardiovascular disease in breast cancer survivors are needed to improve the delivery of survivorship care. Objective: This study aims to assess current cardiovascular risk counseling practices and perceived cardiovascular risk in breast cancer survivors. Methods: Semistructured interviews were conducted from May to December 2021 with breast cancer survivors identified as having a primary care clinician within an academic family medicine center in Charleston, South Carolina. The interview guide and content were developed using the Health Belief Model with a focus on cardiovascular risk behaviors, risk perception, and barriers to risk reduction. Analysis of categorical data was conducted by frequency and quantitative variables by mean and SD. Template analysis was performed for qualitative analysis. Outcome measures included self-reported history of cardiovascular disease, risk perception, and risk behaviors. Results: The average age of participants (n=19) was 54 (SD 7) years; 68\% (13/19) were White and 32\% (6/19) were Black or African American. Of the interviewed women, 90\% (17/19) reported a personal history and 90\% (17/19) reported a family history of cardiovascular disease. Only 53\% (10/19) had previously reported receipt of cardiovascular counseling. Primary care most commonly provided counseling, followed by oncology. Among breast cancer survivors, 32\% (6/19) reported being at increased cardiovascular risk, and 47\% (9/19) were unsure of their relative cardiovascular risk. Factors affecting perceived cardiovascular risk included family history, cancer treatments, cardiovascular diagnoses, and lifestyle factors. Video (15/19, 79\%) and SMS text messaging (13/19, 68\%) were the most highly reported mechanisms through which breast cancer survivors requested to receive additional information and counseling on cardiovascular risk and risk reduction. Commonly reported barriers to risk reduction such as physical activity included time for meal planning and exercise, resources to support dietary and exercise changes, physical limitations, and competing responsibilities. Barriers specific to survivorship status included concerns for immune status during the COVID-19 pandemic, physical limitations associated with cancer treatment, and psychosocial aspects of cancer survivorship. Conclusions: Breast cancer survivors identified that factors associated with their cancer diagnosis and treatment both impacted their cardiovascular risk and introduced additional barriers to risk reduction. Potential strategies to improve counseling and awareness around cardiovascular risk include video and messaging platforms. Further risk reduction strategies should consider the unique challenges of cancer survivorship in delivery and implementation. ", doi="10.2196/71062", url="https://cancer.jmir.org/2025/1/e71062" } @Article{info:doi/10.2196/71948, author="Salustri, Alessandro and Tonti, Giovanni and Pedrizzetti, Gianni and Zhankorazova, Aizhan and Khamitova, Zaukiya and Toktarbay, Bauyrzhan and Jumadilova, Dinara and Khvan, Marina and Galiyeva, Dinara and Bekbossynova, Makhabbat and Mukarov, Murat and Kokoshko, Alexey and Gaipov, Abduzhappar", title="Intradialytic Changes and Prognostic Value of Ventriculo-Arterial Coupling in Patients With End-Stage Renal Disease: Protocol for an Observational Prospective Trial", journal="JMIR Res Protoc", year="2025", month="Jun", day="23", volume="14", pages="e71948", keywords="ventriculo-arterial coupling", keywords="pressure-volume loop", keywords="end-stage renal disease", keywords="hemodialysis", keywords="echocardiography", abstract="Background: The acute effect of hemodialysis (HD) on left ventricular mechanics has been evaluated in several studies; however, their results are not consistent. Eventually, the heart and the arterial system behave as an interconnected system and not as isolated structures; thus, the evaluation of the interaction of cardiac contractility with the arterial system would provide a more comprehensive understanding of cardiovascular function and cardiac energetics. However, there have not been any studies demonstrating changes in terms of volumes, contractility, intraventricular pressure gradient distribution, and vascular properties in response to changes in loading conditions and their impact on the outcome in patients undergoing HD. Recently, a noninvasive method for assessing left ventricular pressure-volume loop and ventriculo-arterial coupling (VAC) from feature-tracking cardiac magnetic resonance or echocardiography has been proposed. We believe that this method allows a comprehensive evaluation of the hemodynamic status of the patients undergoing HD, including the relationships between cardiac function and arterial elastance, and might provide prognostic information. Objective: The primary objective of this study is to evaluate changes in VAC before and after a HD session. The secondary objective is to assess the prognostic value of VAC parameters in predicting adverse outcomes. Methods: A 2D transthoracic echocardiogram will be performed before and after a HD session in patients with end-stage renal disease. We target to enroll 323 patients. Images will be analyzed with advanced software based on speckle-tracking, able to reconstruct the pressure-volume loop. From the pressure-volume loop, arterial (Ea) and ventricular (Ees) elastance will be derived. VAC will be calculated as the Ea/Ees ratio. Patients will be followed up for 18 months. Primary endpoints will be a composite of all causes of death, nonfatal myocardial infarction, and hospitalization due to worsening heart failure. Results: The study received funding in August 2024, with patients' enrollment scheduled to take place from January 1 to June 30, 2025. Data analysis will start in April 2025 and is expected to continue until June 2026. The findings of the study are tentatively planned for publication in the winter of 2027. Conclusions: This study will provide data on the changes in VAC induced by HD and their potential prognostic value. This assessment could be useful for tailoring volume depletion during HD and to improve patients' outcomes. Trial Registration: ClinicalTrials.gov NCT06622928; https://clinicaltrials.gov/study/NCT06622928 International Registered Report Identifier (IRRID): PRR1-10.2196/71948 ", doi="10.2196/71948", url="https://www.researchprotocols.org/2025/1/e71948", url="http://www.ncbi.nlm.nih.gov/pubmed/40550123" } @Article{info:doi/10.2196/68898, author="Alhumaidi, Hamad Norah and Dermawan, Doni and Kamaruzaman, Farhana Hanin and Alotaiq, Nasser", title="The Use of Machine Learning for Analyzing Real-World Data in Disease Prediction and Management: Systematic Review", journal="JMIR Med Inform", year="2025", month="Jun", day="19", volume="13", pages="e68898", keywords="machine learning", keywords="big data", keywords="real-world data", keywords="disease prediction", keywords="health care management", keywords="real-world evidence", keywords="artificial intelligence", keywords="AI", abstract="Background: Machine learning (ML) and big data analytics are rapidly transforming health care, particularly disease prediction, management, and personalized care. With the increasing availability of real-world data (RWD) from diverse sources, such as electronic health records (EHRs), patient registries, and wearable devices, ML techniques present substantial potential to enhance clinical outcomes. Despite this promise, challenges such as data quality, model transparency, generalizability, and integration into clinical practice persist. Objective: This systematic review aims to examine the use of ML for analyzing RWD in disease prediction and management, identifying the most commonly used ML methods, prevalent disease types, study designs, and the sources of real-world evidence (RWE). It also explores the strengths and limitations of current practices, offering insights for future improvements. Methods: A comprehensive search was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines to identify studies using ML techniques for analyzing RWD in disease prediction and management. The search focused on extracting data regarding the ML algorithms applied; disease categories studied; types of study designs (eg, clinical trials and cohort studies); and the sources of RWE, including EHRs, patient registries, and wearable devices. Studies published between 2014 and 2024 were included to ensure the analysis of the most recent advances in the field. Results: This review identified 57 studies that met the inclusion criteria, with a total sample size of >150,000 patients. The most frequently applied ML methods were random forest (n=24, 42\%), logistic regression (n=21, 37\%), and support vector machines (n=18, 32\%). These methods were predominantly used for predictive modeling across disease areas, including cardiovascular diseases (n=19, 33\%), cancer (n=9, 16\%), and neurological disorders (n=6, 11\%). RWE was primarily sourced from EHRs, patient registries, and wearable devices. A substantial portion of studies (n=38, 67\%) focused on improving clinical decision-making, patient stratification, and treatment optimization. Among these studies, 14 (25\%) focused on decision-making; 12 (21\%) on health care outcomes, such as quality of life, recovery rates, and adverse events; and 11 (19\%) on survival prediction, particularly in oncology and chronic diseases. For example, random forest models for cardiovascular disease prediction demonstrated an area under the curve of 0.85 (95\% CI 0.81-0.89), while support vector machine models for cancer prognosis achieved an accuracy of 83\% (P=.04). Despite the promising outcomes, many (n=34, 60\%) studies faced challenges related to data quality, model interpretability, and ensuring generalizability across diverse patient populations. Conclusions: This systematic review highlights the significant potential of ML and big data analytics in health care, especially for improving disease prediction and management. However, to fully realize the benefits of these technologies, future research must focus on addressing the challenges of data quality, enhancing model transparency, and ensuring the broader applicability of ML models across diverse populations and clinical settings. ", doi="10.2196/68898", url="https://medinform.jmir.org/2025/1/e68898" } @Article{info:doi/10.2196/71103, author="Russell, Marta and Cain, Erin and Bazzano, Lydia and De Anda, Ileana and Woo, G. Jessica and Urbina, M. Elaine", title="Collecting at-Home Biometric Measures for Longitudinal Research From the i3C: Feasibility and Acceptability Study", journal="JMIR Hum Factors", year="2025", month="Jun", day="18", volume="12", pages="e71103", keywords="mHealth", keywords="cardiovascular risk factors", keywords="epidemiology", keywords="wearable devices", keywords="feasibility", keywords="acceptability", keywords="biometric", abstract="Background: The use of individual wearable devices or internet-based applications to collect biometric data from research participants is popular, but several devices may be needed to replace a full set of research measurements. Objective: In this study, we assessed the feasibility of a ``Virtual Home Clinic'' within the context of long-term epidemiologic studies. Methods: Participants from 3 study cohorts were recruited. Devices were sent to the home to measure anthropometrics, resting metabolic rate, blood pressure (BP), heart rate (HR), heart rhythm, oxygen saturation, glucose, total cholesterol, physical activity, diet, sleep duration or quality, and arterial stiffness over the course of 1 week. Stool and saliva were also self-collected for microbiome, DNA, and cotinine. Feasibility and acceptability of collecting measurements using home devices were assessed. Results: A total of 134 participants were enrolled (87\% female, 31\% Black; mean age 54.2, SD 8.4 years). Furthermore, 91\% (N=122) performed at least one of the home tests. At least two-thirds of participants were able to complete all of the requested readings for glucose, electrocardiogram, BP, diet record, and resting metabolic rate. The scale that measured weight, body composition, and pulse wave velocity (PWV) was more difficult to use (113/134, 84\% participants recorded at least one weight and 84/134, 63\% recorded a PWV). The device to measure total cholesterol was least successful (32/134, 24\% participants completed all readings, 72/134, 54\% provided at least one result). Return of biospecimens was highly successful (115/134, 86\% returned saliva and 113/134, 84\% returned stool). Of 95 who responded to the user acceptability survey, 38 (40\%) participants preferred home assessment, 36 (38\%) preferred clinic, and 21 (22\%) did not have a preference. The mean user acceptability score across devices for ease of use was 4.3 (SD 1.0), for instructions was 4.5 (SD 0.7), and for time to use was 3.9 (SD 1.1; scale of 1?5, with higher scores indicating greater acceptability). The study team documented several regulatory or IT, connectivity or account, data retrieval, and logistical issues encountered during the study. Conclusions: Despite several complications involved with managing multiple devices and applications, most of the components of the virtual home clinic were reasonably feasible and acceptable to participants. ", doi="10.2196/71103", url="https://humanfactors.jmir.org/2025/1/e71103" } @Article{info:doi/10.2196/71314, author="Lin, Yen-Hung and Chou, Hsu-Wen and Tsai, Sarah and Gomez, Roy", title="Real-World Characteristics and Treatment Patterns of Patients With Transthyretin Amyloid Cardiomyopathy: Protocol for a Multicountry Disease Registry Study", journal="JMIR Res Protoc", year="2025", month="Jun", day="6", volume="14", pages="e71314", keywords="ATTR-CM", keywords="transthyretin amyloid cardiomyopathy", keywords="registry", keywords="real world data", abstract="Background: Transthyretin amyloid cardiomyopathy (ATTR-CM) is a systemic amyloidosis disorder with early clinical manifestations similar to other heart conditions, which complicates its diagnosis and management. The disease's insidious nature and its progression to heart failure emphasize the critical need for enhanced recognition and understanding of its clinical landscape. Objective: This study aimed to understand the natural history and current treatment patterns for managing ATTR-CM in a diverse Asian cohort from Taiwan, Hong Kong, and Malaysia. Methods: This study is a multicenter, noninterventional disease registry that plans to enroll patients diagnosed with ATTR-CM across approximately 17 sites in Taiwan, Hong Kong, and Malaysia. Almost 350 patients with a documented diagnosis of ATTR?CM after June 1, 2019, will be enrolled in the study. Deceased patients will be enrolled without the need for consent in accordance with applicable regulations. Their data will be gathered retrospectively through a 1-time review of their medical records, where permissible. Data related to clinical characteristics, treatment, and outcomes will be collected for each patient during the routine clinical practice while adhering to local standards of care. The end of data collection is planned for at least 12 months after the end of the enrollment period. Results: As of March 16, 2025, ethical approvals for this study have been obtained or are under review at multiple sites across Taiwan, Hong Kong, and Malaysia. The study commenced on October 1, 2024, with the first participant's first visit and so far, 59 patients have been recruited: 35 from National Taiwan University Hospital (Taiwan), 13 from Taipei Veterans General Hospital (Taiwan), 2 from China Medical University Hospital (Taiwan), 2 from Sarawak Heart Center (Malaysia), and 7 from Queen Mary Hospital (Hong Kong). An interim report is scheduled for completion by December 31, 2025. The end of data collection, marked by the last participant's visit, is planned for October 1, 2027, and the final study report is expected to be finalized by June 1, 2028. Once established, the database will serve as a comprehensive resource for analyzing baseline characteristics, treatment patterns, and outcomes in patients with ATTR-CM from diverse health care systems. Conclusions: This research will aid in understanding the demographic, clinical, and therapeutic patterns of ATTR-CM in Taiwan, Hong Kong, and Malaysia. This registry may influence advancements in early detection, diagnosis, and tailored treatment strategies in ATTR-CM. Trial Registration: ClinicalTrials.gov NCT06651073; https://clinicaltrials.gov/study/NCT06651073 International Registered Report Identifier (IRRID): DERR1-10.2196/71314 ", doi="10.2196/71314", url="https://www.researchprotocols.org/2025/1/e71314" } @Article{info:doi/10.2196/68138, author="Amirahmadi, Ali and Etminani, Farzaneh and Bj{\"o}rk, Jonas and Melander, Olle and Ohlsson, Mattias", title="Trajectory-Ordered Objectives for Self-Supervised Representation Learning of Temporal Healthcare Data Using Transformers: Model Development and Evaluation Study", journal="JMIR Med Inform", year="2025", month="Jun", day="4", volume="13", pages="e68138", keywords="patient trajectories", keywords="disease prediction", keywords="representation learning", keywords="masked language mode", keywords="deep learning", keywords="BERT", keywords="electronic health record", keywords="language mode", keywords="transformer", keywords="heart failure", keywords="alzheimer disease", keywords="prolonged health of stay", keywords="effectiveness", keywords="temporal", abstract="Background: The growing availability of electronic health records (EHRs) presents an opportunity to enhance patient care by uncovering hidden health risks and improving informed decisions through advanced deep learning methods. However, modeling EHR sequential data, that is, patient trajectories, is challenging due to the evolving relationships between diagnoses and treatments over time. Significant progress has been achieved using transformers and self-supervised learning. While BERT-inspired models using masked language modeling (MLM) capture EHR context, they often struggle with the complex temporal dynamics of disease progression and interventions. Objective: This study aims to improve the modeling of EHR sequences by addressing the limitations of traditional transformer-based approaches in capturing complex temporal dependencies. Methods: We introduce Trajectory Order Objective BERT (Bidirectional Encoder Representations from Transformers; TOO-BERT), a transformer-based model that advances the MLM pretraining approach by integrating a novel TOO to better learn the complex sequential dependencies between medical events. TOO-Bert enhanced the learned context by MLM by pretraining the model to distinguish ordered sequences of medical codes from permuted ones in a patient trajectory. The TOO is enhanced by a conditional selection process that focus on medical codes or visits that frequently occur together, to further improve contextual understanding and strengthen temporal awareness. We evaluate TOO-BERT on 2 extensive EHR datasets, MIMIC-IV hospitalization records and the Malmo Diet and Cancer Cohort (MDC)---comprising approximately 10 and 8 million medical codes, respectively. TOO-BERT is compared against conventional machine learning methods, a transformer trained from scratch, and a transformer pretrained on MLM in predicting heart failure (HF), Alzheimer disease (AD), and prolonged length of stay (PLS). Results: TOO-BERT outperformed conventional machine learning methods and transformer-based approaches in HF, AD, and PLS prediction across both datasets. In the MDC dataset, TOO-BERT improved HF and AD prediction, increasing area under the receiver operating characteristic curve (AUC) scores from 67.7 and 69.5 with the MLM-pretrained Transformer to 73.9 and 71.9, respectively. In the MIMIC-IV dataset, TOO-BERT enhanced HF and PLS prediction, raising AUC scores from 86.2 and 60.2 with the MLM-pretrained Transformer to 89.8 and 60.4, respectively. Notably, TOO-BERT demonstrated strong performance in HF prediction even with limited fine-tuning data, achieving AUC scores of 0.877 and 0.823, compared to 0.839 and 0.799 for the MLM-pretrained Transformer, when fine-tuned on only 50\% (442/884) and 20\% (176/884) of the training data, respectively. Conclusions: These findings demonstrate the effectiveness of integrating temporal ordering objectives into MLM-pretrained models, enabling deeper insights into the complex temporal relationships inherent in EHR data. Attention analysis further highlights TOO-BERT's capability to capture and represent sophisticated structural patterns within patient trajectories, offering a more nuanced understanding of disease progression. ", doi="10.2196/68138", url="https://medinform.jmir.org/2025/1/e68138", url="http://www.ncbi.nlm.nih.gov/pubmed/40465350" } @Article{info:doi/10.2196/71408, author="Sjoblom, Linnea and Stenbeck, Freja and Trolle Lagerros, Ylva and Hantikainen, Essi and Bonn, E. Stephanie", title="Engagement With a Smartphone-Delivered Dietary Education Intervention and Its Relation to Dietary Intake and Cardiometabolic Risk Markers in People With Type 2 Diabetes: Secondary Analysis of a Randomized Controlled Trial", journal="JMIR Form Res", year="2025", month="May", day="30", volume="9", pages="e71408", keywords="adherence", keywords="dietary change", keywords="diabetes mellitus", keywords="type 2 diabetes mellitus", keywords="healthy diet", keywords="mHealth", keywords="smartphone app", keywords="user engagement", keywords="mobile phone", abstract="Background: Mobile health (mHealth) interventions offer a promising way to support healthy lifestyle habits, but effectiveness depends on user engagement. Maintaining high user engagement in app-based interventions is important, yet challenging. Objective: We aimed to examine the association between user engagement with an app-based dietary education for people with type 2 diabetes and changes in diet quality, dietary intake, and clinical measures. Methods: In this randomized clinical trial, people with type 2 diabetes were recruited within primary care and randomized 1:1 to a 12-week smartphone-delivered app-based dietary education or control group. Participants were followed up after 3, 6, and 12 months. Dietary intake was assessed using a food frequency questionnaire. The control group received the app at the 3-month follow-up. User engagement was analyzed among all participants. Categories of high (100\%), moderate (50\%?99.9\%), and low (<50\%) user engagement were created based on the percentage of activities completed in the app. We used paired t tests to compare mean changes in diet quality, dietary intake, and clinical markers within user engagement groups, and fitted linear regression models to analyze differences in change between groups. Results: Data from 119 participants (60.5\%, 72/119 men) were analyzed. The mean age at baseline was 63.2 (SD 10.3) years and mean BMI was 30.1 (SD 5.1) kg/m2. User engagement was high with an average of 77.1\% of app activities completed. More than half (53.8\%, 64/119) of the users showed high user engagement, 21.8\% (26/119) moderate, and 24.4\% (29/119) low. Directly following the app-based education, a significant difference in change was seen for whole grains ($\beta$=20.4, 95\%CI 0.57?40.3) in participants with high user engagement compared to the low user engagement group who decreased their intake (P=.03). At follow-up after 6 to 9 months after completed education, significant differences in change were seen for fiber, wholegrains, carbohydrates, saturated fat, sodium, and total energy in the moderate compared with the low engagement group, and a significant difference in change was seen for carbohydrates in the high, compared with the low, user engagement group. Conclusions: User engagement was generally high for the smartphone-based dietary education, suggesting that an app targeting dietary habits is feasible to use. Those with higher user engagement seem to maintain healthier dietary behaviours over time, compared to those with low user engagement. Future mHealth interventions should focus on ways to engage those with low interest. Trial Registration: ClinicalTrials.gov NCT03784612; https://www.clinicaltrials.gov/study/NCT03784612 International Registered Report Identifier (IRRID): RR2-10.2196/24422 ", doi="10.2196/71408", url="https://formative.jmir.org/2025/1/e71408" } @Article{info:doi/10.2196/72349, author="Li, Yike and Xiao, Mingyang and Li, Yaqian and Lv, Lulu and Zhang, Shanshan and Liu, Yuhui and Zhang, Juan", title="Machine Learning for the Prediction of Acute Kidney Injury in Critically Ill Patients With Coronary Heart Disease: Algorithm Development and Validation", journal="JMIR Med Inform", year="2025", month="May", day="28", volume="13", pages="e72349", keywords="coronary heart disease", keywords="coronary artery disease", keywords="acute kidney injury", keywords="machine learning", keywords="MIMIC-IV database", abstract="Background: Acute kidney injury (AKI) frequently occurs in critically ill patients with coronary heart disease (CHD), and its development markedly elevates mortality rates and prolongs hospitalization duration. Early AKI prediction is crucial for timely intervention and amelioration of patient outcomes. Objective: This study aimed to develop and verify a clinical prediction model for the occurrence of AKI upon admission in the critically ill population with CHD through machine learning (ML). Methods: Data from the MIMIC-IV (Medical Information Mart for Intensive Care IV) version 2.2 database were gathered and included information about critically ill individuals with CHD in the intensive care unit (ICU). The dataset was randomized into a training set (70\%) and a testing set (30\%). Least absolute shrinkage and selection operator (LASSO) regression was used for feature variable selection. ML models, including logistic regression (LR), decision tree (DT), naive Bayes (NB), random forest (RF), extreme gradient boosting (XGBoost), and support vector machine (SVM), were constructed using 13 variables in the training set. The 6 models were compared in the testing set to identify the best-performing model. Subsequently, the model was assessed using calibration curve analysis and decision curve analysis (DCA). External validation was conducted using data from the Second Affiliated Hospital of Zhengzhou University. Ultimately, the predictive model was interpreted via Shapley Additive Explanation (SHAP) values. Results: In total, 2711 patients with CHD admitted to the ICU were selected, with 1809 (66.7\%) having AKI. XGBoost exhibited the best performance regarding discrimination (area under the receiver operating characteristic curve [AUROC]=0.765, 95\% CI 0.731-0.800), accuracy (0.725), and sensitivity (0.759). External validation using a cohort of 226 patients confirmed the strong generalizability of the XGBoost model (AUROC=0.835, 95\% CI 0.782-0.887). Feature importance analyses derived from SHAP values, DT, RF, and XGBoost consistently identified 5 key predictors associated with the development of AKI: mechanical ventilation, use of antiplatelet agents, age, N-terminal pro--B-type natriuretic peptide (NT-proBNP) levels, and acute physiology score III (APSIII). Conclusions: ML models can serve as reliable tools for forecasting AKI in the critically ill population with CHD. The XGBoost model is highly accurate and may aid doctors in identifying high-risk individuals for early intervention to lower mortality. ", doi="10.2196/72349", url="https://medinform.jmir.org/2025/1/e72349" } @Article{info:doi/10.2196/71726, author="Petch, Jeremy and Tabja Bortesi, Pablo Juan and Sheth, Tej and Natarajan, Madhu and Pinilla-Echeverri, Natalia and Di, Shuang and Bangdiwala, I. Shrikant and Mosleh, Karen and Ibrahim, Omar and Bainey, R. Kevin and Dobranowski, Julian and Becerra, P. Maria and Sonier, Katie and Schwalm, Jon-David", title="Coronary Computed Tomographic Angiography to Optimize the Diagnostic Yield of Invasive Angiography for Low-Risk Patients Screened With Artificial Intelligence: Protocol for the CarDIA-AI Randomized Controlled Trial", journal="JMIR Res Protoc", year="2025", month="May", day="21", volume="14", pages="e71726", keywords="artificial intelligence", keywords="coronary artery disease", keywords="coronary computed tomographic angiography", keywords="clinical decision support", keywords="invasive coronary angiography", abstract="Background: Invasive coronary angiography (ICA) is the gold standard in the diagnosis of coronary artery disease (CAD). Being invasive, it carries rare but serious risks including myocardial infarction, stroke, major bleeding, and death. A large proportion of elective outpatients undergoing ICA have nonobstructive CAD, highlighting the suboptimal use of this test. Coronary computed tomographic angiography (CCTA) is a noninvasive option that provides similar information with less risk and is recommended as a first-line test for patients with low-to-intermediate risk of CAD. Leveraging artificial intelligence (AI) to appropriately direct patients to ICA or CCTA based on the predicted probability of disease may improve the efficiency and safety of diagnostic pathways. Objective: he CarDIA-AI (Coronary computed tomographic angiography to optimize the Diagnostic yield of Invasive Angiography for low-risk patients screened with Artificial Intelligence) study aims to evaluate whether AI-based risk assessment for obstructive CAD implemented within a centralized triage process can optimize the use of ICA in outpatients referred for nonurgent ICA. Methods: CarDIA-AI is a pragmatic, open-label, superior randomized controlled trial involving 2 Canadian cardiac centers. A total of 252 adults referred for elective outpatient ICA will be randomized 1:1 to usual care (directly proceeding to ICA) or to triage using an AI-based decision support tool. The AI-based decision support tool was developed using referral information from over 37,000 patients and uses a light gradient boosting machine model to predict the probability of obstructive CAD based on 42 clinically relevant predictors, including patient referral information, demographic characteristics, risk factors, and medical history. Participants in the intervention arm will have their ICA referral forms and medical charts reviewed, and select details entered into the decision support tool, which recommends CCTA or ICA based on the patient's predicted probability of obstructive CAD. All patients will receive the selected imaging modality within 6 weeks of referral and will be subsequently followed for 90 days. The primary outcome is the proportion of normal or nonobstructive CAD diagnosed via ICA and will be assessed using a 2-sided z test to compare the patients referred for cardiac investigation with normal or nonobstructive CAD diagnosed through ICA between the intervention and control groups. Secondary outcomes include the number of angiograms avoided and the diagnostic yield of ICA. Results: Recruitment began on January 9, 2025, and is expected to conclude in mid to late 2025. As of April 14, 2025, we have enrolled 81 participants. Data analysis will begin once data collection is completed. We expect to submit the results for publication in 2026. Conclusions: CarDIA-AI will be the first randomized controlled trial using AI to optimize patient selection for CCTA versus ICA, potentially improving diagnostic efficiency, avoiding unnecessary complications of ICA, and improving health care resource usage. Trial Registration: ClinicalTrials.gov NCT06648239; https://clinicaltrials.gov/study/NCT06648239/ International Registered Report Identifier (IRRID): DERR1-10.2196/71726 ", doi="10.2196/71726", url="https://www.researchprotocols.org/2025/1/e71726", url="http://www.ncbi.nlm.nih.gov/pubmed/40397500" } @Article{info:doi/10.2196/68066, author="Vu, Thien and Kokubo, Yoshihiro and Inoue, Mai and Yamamoto, Masaki and Mohsen, Attayeb and Martin-Morales, Agustin and Dawadi, Research and Inoue, Takao and Tay, Ting Jie and Yoshizaki, Mari and Watanabe, Naoki and Kuriya, Yuki and Matsumoto, Chisa and Arafa, Ahmed and Nakao, M. Yoko and Kato, Yuka and Teramoto, Masayuki and Araki, Michihiro", title="Machine Learning Model for Predicting Coronary Heart Disease Risk: Development and Validation Using Insights From a Japanese Population--Based Study", journal="JMIR Cardio", year="2025", month="May", day="12", volume="9", pages="e68066", keywords="coronary heart disease", keywords="machine learning", keywords="logistic regression", keywords="random forest", keywords="support vector machine", keywords="Extreme Gradient Boosting", keywords="Light Gradient-Boosting Machine", keywords="Shapley Additive Explanations", keywords="CHD", keywords="SVM", keywords="XGBoost", keywords="LightGBM", keywords="SHAP", abstract="Background: Coronary heart disease (CHD) is a major cause of morbidity and mortality worldwide. Identifying key risk factors is essential for effective risk assessment and prevention. A data-driven approach using machine learning (ML) offers advanced techniques to analyze complex, nonlinear, and high-dimensional datasets, uncovering novel predictors of CHD that go beyond the limitations of traditional models, which rely on predefined variables. Objective: This study aims to evaluate the contribution of various risk factors to CHD, focusing on both established and novel markers using ML techniques. Methods: The study recruited 7672 participants aged 30-84 years from Suita City, Japan, between 1989 and 1999. Over an average of 15 years, participants were monitored for cardiovascular events. A total of 7260 participants and 28 variables were included in the analysis after excluding individuals with missing outcome data and eliminating unnecessary variables. Five ML models---logistic regression, random forest (RF), support vector machine, Extreme Gradient Boosting, and Light Gradient-Boosting Machine---were applied for predicting CHD incidence. Model performance was evaluated using accuracy, sensitivity, specificity, precision, area under the curve, F1-score, calibration curves, observed-to-expected ratios, and decision curve analysis. Additionally, Shapley Additive Explanations (SHAPs) were used to interpret the prediction models and understand the contribution of various risk factors to CHD. Results: Among 7260 participants, 305 (4.2\%) were diagnosed with CHD. The RF model demonstrated the highest performance, with an accuracy of 0.73 (95\% CI 0.64?0.80), sensitivity of 0.74 (95\% CI 0.62?0.84), specificity of 0.72 (95\% CI 0.61?0.83), and an area under the curve of 0.73 (95\% CI 0.65?0.80). RF also showed excellent calibration, with predicted probabilities closely aligning with observed outcomes, and provided substantial net benefit across a range of risk thresholds, as demonstrated by decision curve analysis. SHAP analysis elucidated key predictors of CHD, including the intima-media thickness (IMT\_cMax) of the common carotid artery, blood pressure, lipid profiles (non--high-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides), and estimated glomerular filtration rate. Novel risk factors identified as significant contributors to CHD risk included lower calcium levels, elevated white blood cell counts, and body fat percentage. Furthermore, a protective effect was observed in women, suggesting the potential necessity for gender-specific risk assessment strategies in future cardiovascular health evaluations. Conclusions: We developed a model to predict CHD using ML and applied SHAP methods for interpretation. This approach highlights the multifactor nature of CHD risk evaluation, aiming to support health care professionals in identifying risk factors and formulating effective prevention strategies. ", doi="10.2196/68066", url="https://cardio.jmir.org/2025/1/e68066" } @Article{info:doi/10.2196/69102, author="Nguyen-Huynh, N. Mai and Alexander, Janet and Zhu, Zheng and Meighan, Melissa and Escobar, Gabriel", title="Effects of the National Institutes of Health Stroke Scale and Modified Rankin Scale on Predictive Models of 30-Day Nonelective Readmission and Mortality After Ischemic Stroke: Cohort Study", journal="JMIR Med Inform", year="2025", month="May", day="9", volume="13", pages="e69102", keywords="ischemic stroke", keywords="readmission", keywords="predictive modeling", keywords="mortality", keywords="National Institutes of Health Stroke Scale", keywords="NIHSS", keywords="modified Rankin scale", keywords="mRS", abstract="Background: Patients with stroke have high rates of all-cause readmission and case fatality. Limited information is available on how to predict these outcomes. Objective: We aimed to assess whether adding the initial National Institutes of Health Stroke Scale (NIHSS) score or modified Rankin scale (mRS) score at discharge improved predictive models of 30-day nonelective readmission or 30-day mortality poststroke. Methods: Using a cohort of patients with ischemic stroke in a large multiethnic integrated health care system from June 15, 2018, to April 29, 2020, we tested 2 predictive models for a composite outcome (30-day nonelective readmission or death). The models were based on administrative data (Length of Stay, Acuity, Charlson Comorbidities, Emergency Department Use score; LACE) as well as a comprehensive model (Transition Support Level; TSL). The models, initial NIHSS score, and mRS scores at discharge, were tested independently and in combination with age and sex. We assessed model performance using the area under the receiver operator characteristic (c-statistic), Nagelkerke pseudo-R2, and Brier score. Results: The study cohort included 4843 patients with 5014 stroke hospitalizations. Average age was 71.9 (SD 14) years, 50.6\% (2537/5014) were female, and 52.1\% (2614/5014) were White. Median initial NIHSS score was 4 (IQR 2-8). There were 538 (10.7\%) nonelective readmissions and 150 (3.9\%) deaths within 30 days. The logistic models revealed that the best performing models were TSL (c-statistic=0.69) and TSL plus mRS score at discharge (c-statistic=0.69). Conclusions: We found that neither the initial NIHSS score nor the mRS score at discharge significantly enhanced the predictive ability of the LACE or TSL models. Future efforts at prediction of short-term stroke outcomes will need to incorporate new data elements. ", doi="10.2196/69102", url="https://medinform.jmir.org/2025/1/e69102" } @Article{info:doi/10.2196/56466, author="Sven{\vs}ek, Adrijana and Gosak, Lucija and Lorber, Mateja and {\vS}tiglic, Gregor and Fija{\v c}ko, Nino", title="Review and Comparative Evaluation of Mobile Apps for Cardiovascular Risk Estimation: Usability Evaluation Using mHealth App Usability Questionnaire", journal="JMIR Mhealth Uhealth", year="2025", month="May", day="8", volume="13", pages="e56466", keywords="cardiovascular diseases", keywords="MAUQ", keywords="prognostic models", keywords="mobile applications", keywords="visualization", keywords="PRISMA", abstract="Background: Cardiovascular diseases (CVD) are the leading cause of death and disability worldwide, and their prevention is a major public health priority. Detecting health issues early and assessing risk levels can significantly improve the chances of reducing mortality. Mobile apps can help estimate and manage CVD risks by providing users with personalized feedback, education, and motivation. Incorporating visual analysis into apps is an effective method for educating society. However, the usability evaluation and inclusion of visualization of these apps are often unclear and variable. Objective: The primary objective of this study is to review and compare the usability of existing apps designed to estimate CVD risk using the mHealth App Usability Questionnaire (MAUQ). This is not a traditional usability study involving user interaction design, but rather an assessment of how effectively these applications meet usability standards as defined by the MAUQ. Methods: First, we used predefined criteria to review 16 out of 2238 apps to estimate CVD risk in the Google Play Store and the Apple App Store. Based on the apps' characteristics (ie, developed for health care professionals or patient use) and their functions (single or multiple CVD risk calculators), we conducted a descriptive analysis. Then we also compared the usability of existing apps using the MAUQ and calculated the agreement among 3 expert raters. Results: Most apps used the Framingham Risk Score (8/16, 50\%) and Atherosclerotic Cardiovascular Disease Risk (7/16, 44\%) prognostic models to estimate CVD risk. The app with the highest overall MAUQ score was the MDCalc Medical Calculator (mean 6.76, SD 0.25), and the lowest overall MAUQ score was obtained for the CardioRisk Calculator (mean 3.96, SD 0.21). The app with the highest overall MAUQ score in the ``ease-of-use'' domain was the MDCalc Medical Calculator (mean 7, SD 0); in the domain ``interface and satisfaction,'' it was the MDCalc Medical Calculator (mean 6.67, SD 0.33); and in the domain ``usefulness,'' it was the ASCVD Risk Estimator Plus (mean 6.80, SD 0.32). Conclusions: We found that the Framingham Risk Score is the most widely used prognostic model in apps for estimating CVD risk. The ``ease-of-use'' domain received the highest ratings. While more than half of the apps were suitable for both health care professionals and patients, only a few offered sophisticated visualizations for assessing CVD risk. Less than a quarter of the apps included visualizations, and those that did were single calculators. Our analysis of apps showed that they are an appropriate tool for estimating CVD risk. ", doi="10.2196/56466", url="https://mhealth.jmir.org/2025/1/e56466" } @Article{info:doi/10.2196/67525, author="Park, Jin-Hyun and Jeong, Inyong and Ko, Gang-Jee and Jeong, Seogsong and Lee, Hwamin", title="Development of a Predictive Model for Metabolic Syndrome Using Noninvasive Data and its Cardiovascular Disease Risk Assessments: Multicohort Validation Study", journal="J Med Internet Res", year="2025", month="May", day="2", volume="27", pages="e67525", keywords="metabolic syndrome prediction", keywords="noninvasive data", keywords="clinical interpretable model", keywords="body composition data", keywords="early intervention", abstract="Background: Metabolic syndrome is a cluster of metabolic abnormalities, including obesity, hypertension, dyslipidemia, and insulin resistance, that significantly increase the risk of cardiovascular disease (CVD) and other chronic conditions. Its global prevalence is rising, particularly in aging and urban populations. Traditional screening methods rely on laboratory tests and specialized assessments, which may not be readily accessible in routine primary care and community settings. Limited resources, time constraints, and inconsistent screening practices hinder early identification and intervention. Developing a noninvasive and scalable predictive model could enhance accessibility and improve early detection. Objective: This study aimed to develop and validate a predictive model for metabolic syndrome using noninvasive body composition data. Additionally, we evaluated the model's ability to predict long-term CVD risk, supporting its application in clinical and public health settings for early intervention and preventive strategies. Methods: We developed a machine learning--based predictive model using noninvasive data from two nationally representative cohorts: the Korea National Health and Nutrition Examination Survey (KNHANES) and the Korean Genome and Epidemiology Study. The model was trained using dual-energy x-ray absorptiometry data from KNHANES (2008-2011) and validated internally with bioelectrical impedance analysis data from KNHANES 2022. External validation was conducted using Korean Genome and Epidemiology Study follow-up datasets. Five machine learning algorithms were compared, and the best-performing model was selected based on the area under the receiver operating characteristic curve. Cox proportional hazards regression was used to assess the model's ability to predict long-term CVD risk. Results: The model demonstrated strong predictive performance across validation cohorts. Area under the receiver operating characteristic curve values for metabolic syndrome prediction ranged from 0.8338 to 0.8447 in internal validation, 0.8066 to 0.8138 in external validation 1, and 0.8039 to 0.8123 in external validation 2. The model's predictions were significantly associated with future cardiovascular risk, with Cox regression analysis indicating that individuals classified as having metabolic syndrome had a 1.51-fold higher risk of developing CVD (hazard ratio 1.51, 95\% CI 1.32-1.73; P<.001). The ability to predict long-term CVD risk highlights the potential utility of this model for guiding early interventions. Conclusions: This study developed a noninvasive predictive model for metabolic syndrome with strong performance across diverse validation cohorts. By enabling early risk identification without laboratory tests, the model enhances accessibility in primary care and large-scale screenings. Its ability to predict long-term CVD risk supports proactive intervention strategies, potentially reducing the burden of cardiometabolic diseases. Further research should refine the model with additional clinical factors and broader population validation to maximize its clinical impact. ", doi="10.2196/67525", url="https://www.jmir.org/2025/1/e67525" } @Article{info:doi/10.2196/70587, author="Wang, Jun and Zhu, Jiajun and Li, Hui and Wu, Shili and Li, Siyang and Yao, Zhuoya and Zhu, Tongjian and Tang, Bi and Tang, Shengxing and Liu, Jinjun", title="Multimodal Visualization and Explainable Machine Learning--Driven Markers Enable Early Identification and Prognosis Prediction for Symptomatic Aortic Stenosis and Heart Failure With Preserved Ejection Fraction After Transcatheter Aortic Valve Replacement: Multicenter Cohort Study", journal="J Med Internet Res", year="2025", month="May", day="1", volume="27", pages="e70587", keywords="machine learning", keywords="interpretable models", keywords="heart failure with preserved ejection fraction", keywords="symptomatic aortic stenosis", keywords="transcatheter aortic valve replacement", keywords="major adverse cardiovascular and cerebrovascular events.", abstract="Background: Currently, there is a paucity of literature addressing personalized risk stratification using multimodal data in patients with symptomatic aortic stenosis and heart failure with preserved ejection fraction (HFpEF) following transcatheter aortic valve replacement (TAVR). Objective: This study aimed to enhance the performance of risk assessment models in this patient population by developing a predictive model for adverse outcomes using various machine learning (ML) techniques. Methods: This multicenter cohort study included 326 patients diagnosed with severe AS and HFpEF who underwent TAVR between January 2017 and December 2023. Patients were allocated to training (n=195) and independent validation (n=131) sets based on hospital affiliation. A dual-phase feature selection process, combining least absolute shrinkage and selection operator logistic regression and the Boruta algorithm, was used to identify relevant variables from the multimodal dataset. A total of 5 ML model-decision trees, K-nearest neighbors, random forest, support vector machine, and extreme gradient boosting were used to construct a visualization and explainable predictive framework to elucidate model decision-making processes. Results: The primary features identified included age, N-terminal pro-brain natriuretic peptide, fasting blood glucose, triglyceride/high-density lipoprotein cholesterol ratio, triglyceride glucose index, triglyceride glucose-BMI index, atherogenic index of plasma index, and Apolipoprotein B. Among the 5 models, the support vector machine demonstrated the best predictive performance for major adverse cardiovascular and cerebrovascular events in patients with severe AS and HFpEF following TAVR, achieving an area under the curve of 0.756 (95\% CI 0.631-0.881) in the independent validation set. The model exhibited good calibration and robust predictive power in both training and validation sets and demonstrated the highest net benefit in decision curve analysis compared to other models. To extract significant variables influencing the algorithm and ensure model appropriateness, we interpreted cohort and personalized model predictions using Shapley Additive Explanations values. Conclusions: Our ML-based multimodal model, incorporating 8 readily accessible predictors, demonstrated robust predictive capability for 12 months of major adverse cardiovascular and cerebrovascular events risk. This model can be used to identify high-risk individuals with AS and HFpEF following TAVR, potentially aiding in risk stratification and personalized treatment strategies. ", doi="10.2196/70587", url="https://www.jmir.org/2025/1/e70587", url="http://www.ncbi.nlm.nih.gov/pubmed/40310672" } @Article{info:doi/10.2196/68929, author="Yang, Li-Tan and Wu, Chi-Han and Lee, Jen-Kuang and Wang, Wei-Jyun and Chen, Ying-Hsien and Huang, Ching-Chang and Hung, Chi-Sheng and Chiang, Kuang-Chien and Ho, Yi-Lwun and Wu, Hui-Wen", title="Effects of a Cloud-Based Synchronous Telehealth Program on Valvular Regurgitation Regression: Retrospective Study", journal="J Med Internet Res", year="2025", month="Apr", day="23", volume="27", pages="e68929", keywords="mitral regurgitation", keywords="tricuspid regurgitation", keywords="telehealth", keywords="telemedicine", keywords="cardiac remodeling", abstract="Background: Telemedicine has been associated with better cardiovascular outcomes, but its effects on the regression of mitral regurgitation (MR) and tricuspid regurgitation (TR) remain unknown. Objective: This study aimed to evaluate whether telemedicine could facilitate the regression of MR and TR compared to usual care and whether it was associated with better survival. Methods: This retrospective cohort study enrolled consecutive patients with moderate or greater MR or TR from 2010 through 2020, excluding those with concomitant aortic stenosis, aortic regurgitation, or mitral stenosis greater than mild severity. All patients underwent follow-up transthoracic echocardiography (TTE) at least 3 months apart. Patients receiving telehealth services for at least two weeks within 90 days of baseline TTE were categorized as the telehealth group; the remainder constituted the nontelehealth group. Telemedicine participants transmitted daily biometric data---blood pressure, pulse rate, blood glucose, electrocardiogram, and oxygen saturation---to a cloud-based platform for timely monitoring. Experienced case managers regularly contacted patients and initiated immediate action for concerning measurements. The primary endpoint was MR or TR regression from ?moderate to