Computer methods and programs in biomedicine最新文献

{"title":"An AI-based microsimulation for predicting health outcomes among people experiencing homelessness.","authors":"Antonio Blasco-Calafat, Vicent Blanes-Selva, Ascensión Doñate-Martínez, Tobias Fragner, Tamara Alhambra-Borrás, Julia Gawronska, Maria Moudatsou, Ioanna Tabaki, Katerina Belogianni, Pania Karnaki, Miguel Rico Varadé, Rosa Gómez Trenado, Jaime Barrio-Cortes, Lee Smith, Alejandro Gil-Salmeron, Igor Grabovac, Juan M García-Gómez","doi":"10.1016/j.cmpb.2025.109112","DOIUrl":"https://doi.org/10.1016/j.cmpb.2025.109112","url":null,"abstract":"<p><strong>Background and objective: </strong>People experiencing homelessness (PEH) face higher cancer risk due to social exclusion, housing and limited access to healthcare. This study proposes a microsimulation model using machine learning (ML) to predict the effect of quality of life, healthcare utilisation and empowerment at the end of the intervention under the Health Navigator Model, enabling cost-effective resource allocation and identifying high-risk subgroups.</p><p><strong>Materials & methods: </strong>We used data from 652 PEH recruited in four European countries (June 2022 - November 2023); 255 completed an 18-month Health Navigator Model programme. Standardised questionnaires were administered at baseline, four weeks and post-intervention. A modular ML microsimulation was built that (1) creates a constraint-based synthetic cohort, (2) estimates outcome changes by matching each simulated case to real program completers, and (3) sums those differences to gauge the intervention's impact. Multiple ML techniques were tested to keep the synthetic sample true to the original and to improve effect-size predictions.</p><p><strong>Results: </strong>CTGAN generated the most realistic synthetic baseline (propensity score = 0.152; 95 % CI 0.148-0.162), markedly outperforming univariant, multivariant and SMOTE approaches (> 0.21). Regression models reproduced most numerical outcomes with good fidelity (e.g., EQ-5D-5L MAE = 0.10 on a 0-1 scale; Health-Rating MAE = 10 on a 0-100 scale), while categorical outcomes were predicted within roughly one category. Binary classifiers yielded F1-scores of 0.58 for smoking status and 0.64 for programme adherence. An online demonstrator (https://epione.upv.es) visualises the process.</p><p><strong>Conclusion: </strong>The proposed ML-based microsimulation generates realistic PEH profiles and projects intervention outcomes, providing a flexible, evidence-driven tool to optimise cancer-prevention strategies for PEH supporting evidence-based decision-making and optimise resource allocation, enhancing intervention outcomes by predicting the intervention before implementation.</p>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"109112"},"PeriodicalIF":4.8,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145291424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing post-induction hypotension prediction based on exemplar learning with crossover restart strategy driven feature selection.","authors":"Liufang Sheng, Shenghui Yu, Ke Ding, Ke Yan, Qianfeng Yu, Lei Shi, Lei Li, Ali Asghar Heidari, Huiling Chen, Junping Chen","doi":"10.1016/j.cmpb.2025.109108","DOIUrl":"https://doi.org/10.1016/j.cmpb.2025.109108","url":null,"abstract":"<p><strong>Background and objective: </strong>Post-induction hypotension (PIH), primarily resulting from the vasodilatory effects and reduced cardiac output induced by anesthetic agents, is widespread among patients with pre-existing cardiovascular conditions or those who have experienced suboptimal fluid management. This condition can lead to inadequate perfusion of critical organs such as the brain and heart, increasing the risk of lengthy postoperative recovery, complications, and mortality. Therefore, early identification and prediction of PIH are crucial for improving postoperative management and patient outcomes.</p><p><strong>Methods: </strong>This study utilized data from 440 elderly patients experiencing elective surgery under general anesthesia at the Ningbo University Affiliated People's Hospital. Patients were categorized into PIH and non-PIH groups based on their mean arterial pressure at the time of induction. To predict PIH, the study developed a machine-learning model named bECRIME-SVM. The model employed an exemplar learning strategy enhanced by a crossover restart strategy within the rime optimization algorithm (ECRIME) to select optimal feature subsets. These subsets were then evaluated using a support vector machine (SVM) to assess their predictive efficacy for PIH.</p><p><strong>Results: </strong>The bECRIME-SVM model demonstrated strong performance on the PIH dataset, achieving a prediction accuracy of 84.100 % and a specificity of 85.287 %. Comparative analysis with other models from the CEC 2017 benchmark functions confirmed the superior optimization capability and convergence accuracy of the ECRIME algorithm. The model also identified several key predictive features, including diabetes, drinking history, atropine, β-blockers, total cholesterol, and pre-induction systolic blood pressure.</p><p><strong>Conclusions: </strong>The bECRIME-SVM model provides a valuable tool for the clinical prediction of PIH, with high accuracy and specificity. Identifying significant predictive features offers essential insights for the early detection and management of PIH, ultimately contributing to improved postoperative outcomes for patients undergoing general anesthesia.</p>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"109108"},"PeriodicalIF":4.8,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145312533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Let XAI generate reliability metadata, not medical explanations","authors":"Federico Cabitza ,&nbsp;Enea Parimbelli","doi":"10.1016/j.cmpb.2025.109090","DOIUrl":"10.1016/j.cmpb.2025.109090","url":null,"abstract":"<div><div>As AI becomes increasingly embedded in medical practice, the call for explainability – commonly framed as eXplainable AI (XAI) – has grown, especially under regulatory pressures. However, conventional XAI approaches misunderstand clinical decision-making by focusing on post-hoc explanations rather than actionable cues. This letter argues that to calibrate trust in AI recommendations, physicians’ primary need is not for conventional post-hoc explanations, but for “<em>reliability metadata</em>”: a set of both marginal and instance-specific indicators that facilitate the assessment of the reliability of each individual advice given. We propose shifting the focus from generating static explanations to providing actionable cues – such as calibrated confidence scores, out-of-distribution alerts, and relevant reference cases – that support adaptive reliance and mitigate automation bias. By reframing XAI as <em>eXtended and eXplorable AI</em>, we emphasize interaction, uncertainty transparency, and clinical relevance over explanations per se. This perspective encourages AI design that aligns with real-world medical cognition, promotes reflective engagement, and supports safer, more effective decision-making.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109090"},"PeriodicalIF":4.8,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145263408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predictive model of Ki67 expression level in osteosarcoma based on weakly supervised segmentation and multi-type feature fusion","authors":"Qi Wang ,&nbsp;Qun Ma ,&nbsp;Xiuyan Li ,&nbsp;Siqi Ben ,&nbsp;Jun Xue ,&nbsp;Tianrui Shang ,&nbsp;Xiaoxuan Jing ,&nbsp;Aidong Liu","doi":"10.1016/j.cmpb.2025.109098","DOIUrl":"10.1016/j.cmpb.2025.109098","url":null,"abstract":"<div><h3>Background and objective</h3><div>Osteosarcoma is a highly malignant bone tumor that occurs primarily in children and adolescents. Ki67 protein expression level (detected through immunohistochemistry) is an important indicator for assessing tumor proliferative activity. This study aims to develop an efficient and low-cost artificial intelligence model to predict Ki67 expression levels from pathological images.</div></div><div><h3>Methods</h3><div>73 hematoxylin and eosin-stained (H&amp;E) whole slide images (WSIs) of osteosarcoma specimens were analyzed. Tumor regions were segmented using weakly supervised learning, followed by extraction of 215 nuclear features including shape, texture, spatial and topological features through the Hover-Net network. Feature selection was performed using five methods: least absolute shrinkage and selection operator (LASSO), mutual information (MI), recursive feature elimination (RFE), Wilcoxon rank sum test (WRST), and extreme gradient boosting (XGBoost), with the top 5 features selected from each method. These features were subsequently integrated with 8 machine learning classifiers: adaptive boosting (AdaBoost), balanced random forest (BalancedRF), k-nearest neighbors (KNN), light gradient boosting machine (LightGBM), multilayer perceptron (MLP), quadratic discriminant analysis (QDA), random forest (RF), and support vector machine (SVM) to determine the optimal hybrid model.</div></div><div><h3>Results</h3><div>By combining 5 key features with 8 machine learning classifiers, we selected the optimal hybrid model (XGBoost+SVM). This model demonstrated the best performance in accuracy (0.767 ± 0.018), recall (0.872 ± 0.036), F1-score (0.800 ± 0.012), and receiver operating characteristic-area under curve (ROC-AUC) (0.884 ± 0.045). The model showed both high accuracy and high sensitivity in Ki67 detection.</div></div><div><h3>Conclusion</h3><div>Our model provides an automated and reliable solution for osteosarcoma Ki67 assessment, reducing dependence on traditional immunohistochemistry. Its excellent performance indicates strong potential for clinical translation.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109098"},"PeriodicalIF":4.8,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145257809","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Semi-supervised deep matrix factorization model for clustering multi-omics data.","authors":"Khanh Luong, Nirav Joshi, Richi Nayak","doi":"10.1016/j.cmpb.2025.109094","DOIUrl":"https://doi.org/10.1016/j.cmpb.2025.109094","url":null,"abstract":"<p><strong>Background and objective: </strong>Multi-omics data are inherently high-dimensional, sparse, and noisy, posing significant challenges for clustering and integration. Conventional clustering and linear dimensionality reduction methods often fail to handle noise effectively or provide interpretability, while standard non-negative matrix factorization approaches are too shallow to capture non-linear patterns. Multi-view non-negative matrix factorization enables integration of complementary views, but it remains primarily unsupervised and seldom leverages available label information.</p><p><strong>Methods: </strong>We propose SSD-MO, a Semi-Supervised Deep Non-Negative Matrix Factorization model for Multi-Omics Data, designed to address these challenges by leveraging both labelled and unlabelled samples for enhanced data integration and clustering performance. SSD-MO combines semi-supervised learning with a multi-layer deep factorization framework, preserving local geometric structure and incorporating orthogonal and diversity constraints. Its effectiveness was validated on six multi-omics datasets from The Cancer Genome Atlas, using evaluation metrics such as clustering accuracy, normalized mutual information, and F-scores.</p><p><strong>Results: </strong>SSD-MO significantly improved clustering accuracy, achieving an increase in F-score by 9%-24% compared to unsupervised baselines and 7%-20% over semi-supervised benchmarks. Precision (64%-73%) and Recall (70%-88%) values further demonstrated its robust performance across datasets.</p><p><strong>Conclusion: </strong>This method provides a robust framework for multi-omics data integration and holds promise for applications in genomics and precision medicine.</p>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"109094"},"PeriodicalIF":4.8,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145291451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrative analysis of RNA modification-related gene PUS7 in diagnosis, prognosis, and tumor microenvironment of hepatocellular carcinoma.","authors":"Lin Chen, Man Wang, Meng Pan, Qiang Wang, Hong Zheng, Huimin Li, Xiangqian Guo","doi":"10.1016/j.cmpb.2025.109114","DOIUrl":"https://doi.org/10.1016/j.cmpb.2025.109114","url":null,"abstract":"<p><strong>Background: </strong>Hepatocellular carcinoma is a highly aggressive cancer with a poor prognosis. RNA modifications play critical roles in regulating various biological functions in cancers. However, the involvement of RNA modification-related pseudouridine synthase (PUS) genes in hepatocellular carcinoma remains unclear.</p><p><strong>Methods: </strong>To explore the diagnostic and prognostic significance of PUS genes in hepatocellular carcinoma, we analysed the transcriptomic and clinical data from The Cancer Genome Atlas (TCGA). Differential analysis and ROC analysis were conducted to evaluate the diagnostic potential of PUS genes. Protein expression levels of PUS1 and PUS7 were examined using The Human Protein Atlas (HPA) database, and the overexpression of PUS7 was further validated by immunohistochemistry staining. Univariate and multivariate COX regression analyses were performed to analyze the prognostic value of PUS7. Nomograms were built based on COX regression analytic results to predict patient survival. The tumor immune microenvironment was characterized using TIMER2.0, EPIC, and xCell algorithms to estimate the immune cell infiltration.</p><p><strong>Results: </strong>The results showed that PUS7 mRNA and protein levels are significantly over-expressed in hepatocellular carcinoma tissues compared to normal tissues. PUS7 overexpression is associated with poor prognosis and may serve as an independent prognostic factor for overall survival in hepatocellular carcinoma. Functional enrichment analysis indicated that PUS7 is involved in key oncogenic pathways, including cell cycle, DNA replication, homologous recombination, oocyte meiosis, mismatch repair, and spliceosome signaling pathways. Immune microenvironment analysis revealed a significant correlation between PUS7 expression and immune cell infiltration in hepatocellular carcinoma, suggesting a potential role in modulating tumor immunity.</p><p><strong>Conclusion: </strong>Our findings suggest that PUS7 is a novel and promising biomarker for diagnosis and prognosis of hepatocellular carcinoma. This study provides preliminary insights into the functional roles and underlying mechanisms of PUS7 in HCC progression, offering theoretical support for improving patient outcomes and identifying potential therapeutic targets.</p>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"109114"},"PeriodicalIF":4.8,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145291486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advances in digital health: Multimodal intelligence and translational impact","authors":"Arkadiusz Gertych ,&nbsp;Massimo Salvi ,&nbsp;Massimo Mischi","doi":"10.1016/j.cmpb.2025.109111","DOIUrl":"10.1016/j.cmpb.2025.109111","url":null,"abstract":"","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109111"},"PeriodicalIF":4.8,"publicationDate":"2025-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145263403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Detection-driven two-stage framework for intraoperative ROSE WSI classification","authors":"Yingjiao Deng ,&nbsp;Qing Zhang ,&nbsp;Chunhua Zhou ,&nbsp;Lili Gao ,&nbsp;Xianzheng Qin ,&nbsp;Hui Lu ,&nbsp;Jiansheng Wang ,&nbsp;Li Sun ,&nbsp;Yan Wang ,&nbsp;Duowu Zou ,&nbsp;Hongkai Xiong ,&nbsp;Qingli Li","doi":"10.1016/j.cmpb.2025.109084","DOIUrl":"10.1016/j.cmpb.2025.109084","url":null,"abstract":"<div><h3>Background and objective:</h3><div>Solid pancreatic lesions (SPLs) represent one of the most lethal forms of gastrointestinal malignancies, and Rapid on-site evaluation (ROSE) serves as an important component of intraoperative diagnosis. However, efficient and accurate ROSE slide interpretation remains challenging due to the gigapixel scale of whole-slide images, sparse distribution of diagnostically relevant regions, and the need for real-time feedback.</div></div><div><h3>Methods:</h3><div>To address challenges, we propose a novel two-stage framework for fast and precise ROSE WSI classification, following the clinical diagnostic workflow of cytopathologists. In the first stage, we design a lightweight Transformer-based object detection network named as RoF DETR, which detects key cell clusters at 5x magnification. To further enhance detection performance, we incorporate domain-specific medical foundation model features and design a multi-scale feature fusion module for effective feature extraction. In the second stage, we design a prototype-guided multiple instance learning network (PG-MIL) based on pseudo-bag augmentation for 20x magnification patch extraction, improving feature discrimination and robustness under class imbalance.</div></div><div><h3>Results:</h3><div>For comprehensive evaluation, we establish a dedicated ROSE WSI dataset and a cell cluster detection dataset. Our method achieves an <span><math><mrow><mi>A</mi><mi>P</mi><mi>@</mi><mn>0</mn><mo>.</mo><mn>5</mn></mrow></math></span> of 0.482 in cell cluster detection and an AUC of 92.36% in WSI-level classification. Compared to conventional WSI-level classification pipelines, the proposed framework reduces computational overhead by approximately 100<span><math><mo>×</mo></math></span> and halves the inference time.</div></div><div><h3>Conclusion:</h3><div>The proposed framework provides a scalable and efficient solution for rapid cytological assessment of ROSE slides, showing potential to support real-time intraoperative decision-making in clinical workflows.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109084"},"PeriodicalIF":4.8,"publicationDate":"2025-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145250211","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BrainOSM: Outlier screening for multi-view functional brain network analysis","authors":"Guiliang Guo ,&nbsp;Guangqi Wen ,&nbsp;Lingwen Liu ,&nbsp;Ruoxian Song ,&nbsp;Peng Cao ,&nbsp;Jinzhu Yang ,&nbsp;Osmar R. Zaiane","doi":"10.1016/j.cmpb.2025.109092","DOIUrl":"10.1016/j.cmpb.2025.109092","url":null,"abstract":"<div><h3>Purpose:</h3><div>Identifying biomarkers for mental diseases is vital for understanding their underlying mechanisms, facilitating early diagnosis, and enabling more personalized treatment strategies. In this study, we focus on diagnosing autism spectrum disorder (ASD) and alzheimer’s disease (AD) by analyzing functional brain networks (FBNs), which are represented as graphs capturing the functional connectivity patterns of the brain. The primary challenges in modeling FBNs for this disorder stem from two key issues: (i) the heterogeneity among graphs, and (ii) the disease-unrelated information within graphs.</div></div><div><h3>Method:</h3><div>We introduce a two-stage framework, BrainOSM, which combines outlier screening in datasets with a multi-view graph pooling module for enhanced graph classification. Specifically, the first stage employs progressive uncertainty-based outlier screening to reduce the interference of inter-graph heterogeneity. The second stage integrates multi-graph pooling, multi-view learning, and prior subnetwork regularization to refine graph structures, effectively tackling the challenge of disease-unrelated information within graphs.</div></div><div><h3>Results:</h3><div>To validate the effectiveness of our method, we assess its performance on two public datasets: the Autism Brain Imaging Data Exchange (ABIDE) dataset and the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. On the ABIDE dataset, BrainOSM achieved an average accuracy of 70.23% and an AUC of 70.42%, corresponding to improvements of 8.55% and 7.74% over the traditional GCN method. On the ADNI dataset, it reached an average accuracy of 82.29% and an AUC of 83.23%, showing gains of 8.97% and 11.78%, respectively. Our code is publicly available at <span><span>https://github.com/guoguiliang111/BrainOSM</span><svg><path></path></svg></span>.</div></div><div><h3>Conclusion:</h3><div>Our extensive experiments confirm the generalizability and the effectiveness of BrainOSM for mental disease classification. Visual analyses further demonstrate that the model effectively identifies subnetworks associated with mental diseases, highlighting its potential for clinical interpretation. Moreover, our findings indicate that outlier screening plays a crucial role in improving classification accuracy when dealing with heterogeneous datasets.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109092"},"PeriodicalIF":4.8,"publicationDate":"2025-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145263406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Precision at heart: An IoT-based vertical federated learning approach for heterogeneous data-driven cardiovascular disease risk prediction","authors":"Sulfikar Shajimon ,&nbsp;Raj Mani Shukla ,&nbsp;Amar Nath Patra","doi":"10.1016/j.cmpb.2025.109079","DOIUrl":"10.1016/j.cmpb.2025.109079","url":null,"abstract":"<div><h3>Background:</h3><div>Cardiovascular disease (CVD) seriously threatens individual health, highlighting the importance of early detection and proactive mitigation. With advances in consumer electronics such as wearables and IoT, an opportunity exists to enhance CVD prediction for users. Machine learning (ML) has been widely used for predicting CVD risk (high/low) based on various factors and is a critical area of healthcare research. However, sharing data needed to predict CVD with machine learning models is challenging due to privacy concerns. Federated learning (FL) enables distributed training of ML models without sharing raw data. However, it requires that all training features be available to all clients.</div></div><div><h3>Methods:</h3><div>To address this problem, we propose a method based on vertical federated learning (VFL) designed for consumer electronics platforms. The proposed method trains the Neural Network (NN) model in a distributed manner in which different parties hold different data features. In this work, each party maintains a portion of separate data features, performs calculations on them locally, and then transfers only the necessary information to train a NN model jointly. We employ the proposed method for different use cases where the dataset features are distributed between: (i) the patient and the hospital (2-splits); (ii) the patient, the doctor, and the laboratory (3-splits); and (iii) the patient, the doctor, the Electrocardiogram (ECG) center, and the laboratory (4-splits).</div></div><div><h3>Results:</h3><div>Using a realistic dataset publicly available, we test the proposed methodology, which gives around 90% accuracy, precision, recall, and f-score. It also does not need clients to possess the same features as compared to traditional Federated Learning.</div></div><div><h3>Conclusion:</h3><div>This paper has an impact on the healthcare sector, where user data privacy is of utmost concern. This advancement will improve the ability of the healthcare sector to diagnose various diseases and contribute to the field of AI by improving distributed AI algorithms.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"273 ","pages":"Article 109079"},"PeriodicalIF":4.8,"publicationDate":"2025-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145257796","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}