Computers in biology and medicine最新文献

{"title":"A computational model of tsDCS effects in SOD1 mice: from MRI-based design to validation","authors":"L. de Oliveira Pires ,&nbsp;B. Wasicki ,&nbsp;A. Abaei ,&nbsp;J. Scekic-Zahirovic ,&nbsp;F. Roselli ,&nbsp;S. Fernandes ,&nbsp;M. Bączyk","doi":"10.1016/j.compbiomed.2025.111082","DOIUrl":"10.1016/j.compbiomed.2025.111082","url":null,"abstract":"<div><div>During trans-spinal direct current stimulation (tsDCS) the transmembrane potential of neurons is modified by an electric field (EF) induced due to externally applied direct current (DC). The resultant functional effects are being harnessed in the treatment of various neurological conditions; however, the fundamental mechanisms of action underlying tsDCS remain unclear. This ambiguity is largely attributed to the limited knowledge of the geometrical constraints of the EF in the polarized spinal regions. It is, then, essential to develop tools that enable researchers to plan tsDCS approaches in a controlled and systematic manner, ensuring the reproducibility of stimulation effects at spinal targets.</div><div>With this paper, we aim to provide a comprehensive computational model of tsDCS intervention in mice to support further fundamental research in this area. Our model was constructed using high-resolution MRI scans of C57/B6 mice, which were segmented and reconstructed into a realistic mouse computational model. <em>In vivo</em> electrophysiological measurements of voltage gradients in SOD1 G93A mice were used to validate our model predictions in real-life scenarios. In both the modeling and <em>in vivo</em> studies, we employed a rostrocaudal arrangement of DC electrodes to replicate stimulation parameters that have proven effective for modulating murine spinal circuits.</div><div>Both the computational and <em>in vivo</em> approaches yielded highly consistent results, with EF parameters primarily influenced by the distance between the target site and the tsDCS electrodes. We conclude that this developed model offers high accuracy in EF distribution and can significantly substantiate basic research in tsDCS.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111082"},"PeriodicalIF":6.3,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118162","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":"On the dynamics and mitigation of SARS-CoV-2 outbreak in India via a fractional modelling with the Liouville–Caputo derivative","authors":"Asim Patra ,&nbsp;Mustafa Inc","doi":"10.1016/j.compbiomed.2025.111019","DOIUrl":"10.1016/j.compbiomed.2025.111019","url":null,"abstract":"<div><div>The dawn of the year 2019 had brought in a dangerous demon called novel coronavirus which wrecked havoc on the entire world. India has observed the very first case of SARS-CoV-2 infection on 30 January, 2020 and till date as of September, 2025, there are 70,47,53,890 total cases out of which 70,10,681 deaths were recorded. Due to this reason there had been an urge of studying the dynamics of this deadly disease and mathematical models are effective way to have an analytical study of the disease outbreak and control. Hence, in this work, an epidemic modelling has been efficiently formulated as a Susceptible(<span><math><mi>S</mi></math></span>)-Exposed(<span><math><mi>E</mi></math></span>)-Infected(<span><math><mi>I</mi></math></span>)-Asymptomatically infected(<span><math><mi>A</mi></math></span>)-Quarantined(<span><math><mi>Q</mi></math></span>)-Hospitalized(<span><math><mi>H</mi></math></span>)-Recovered(<span><math><mi>R</mi></math></span>) (<span><math><mi>S</mi></math></span> <span><math><mi>E</mi></math></span> <span><math><mi>I</mi></math></span> <span><math><mi>A</mi></math></span> <span><math><mi>Q</mi></math></span> <span><math><mi>H</mi></math></span> <span><math><mi>R</mi></math></span>) model, where the derivative is applied in the Liouville–Caputo (L-C) fractional sense. An analysis has been done regarding the transmission dynamics, causes, control, preventive measures like the effect of quarantine, self-isolation, environmental impacts and other initiatives that has been taken by the Indian government to alleviate the spread of the communicable virus. The data available regarding the infected population of India from 27th July, 2020 to 26th July, 2021 has been taken into consideration (Figure 2) for analysis and, in accordance with that, several parameter values of the fractional model have been fitted. Furthermore, the local stability and global stability analysis have been conducted for the pandemic equilibrium state. In addition to that, based on the sensitivity analysis along with the dynamics of the threshold values like the reproduction ratio <span><math><mrow><msub><mrow><mi>R</mi></mrow><mrow><mn>0</mn></mrow></msub><mo>≈</mo><mn>2</mn><mo>.</mo><mn>589</mn></mrow></math></span>, we figure out the efficiency of the preventive strategies, which forecast the outbreaks in future and potential of the control measures of the disease. Apart from that, an effective numerical technique has been employed to simulate the model and the resulting numerical values have been further tabulated and depicted graphically.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111019"},"PeriodicalIF":6.3,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118163","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":"Enhanced shear wave attenuation estimation with expanded bandwidth in viscoelastic media","authors":"Piotr Kijanka,&nbsp;Ramin Almasi","doi":"10.1016/j.compbiomed.2025.111042","DOIUrl":"10.1016/j.compbiomed.2025.111042","url":null,"abstract":"<div><div>Ultrasound shear wave elastography (SWE) is widely employed to differentiate healthy from pathological tissues based on their viscoelastic mechanical properties. Although elasticity has traditionally been the main focus, tissue viscosity also plays a crucial role in characterizing mechanical behavior using SWE. Numerous methods have been developed for estimating the viscosity, with rheological model-based approaches being the most widely used. However, model-free techniques are gaining increasing attention, as they do not impose a predetermined relationship between wave velocity and attenuation, offering greater flexibility in capturing complex tissue viscoelastic behavior. In this study, we propose a novel method for calculating the shear wave attenuation. The approach, termed SAGA-ST, integrates a Super-Gaussian window-based Stockwell transform with slant frequency–wavenumber (f-k) analysis and leverages the full width at half-maximum of the f-k spectrum. The SAGA-ST method was first evaluated using analytical phantom data in tissue-mimicking viscoelastic media. To further assess its robustness, we applied the method to analytical shear wave motion data corrupted with varying levels of additive white Gaussian noise. Experimental validation was also performed using data from custom-designed tissue-mimicking phantoms and <em>ex vivo</em> bovine liver sample. We compared the performance of SAGA-ST with two existing techniques: the two-dimensional Fourier transform (2D-FT) and the generalized Stockwell transform-based method (GST-SFK). The proposed SAGA-ST method consistently demonstrates superior performance in analytical phantoms, exhibiting lower median attenuation bias and a smaller interquartile range. This is further supported by data points predominantly falling within the <span><math><mrow><mo>±</mo><mn>5</mn><mtext>%</mtext></mrow></math></span> acceptable bias region and a median attenuation bias of less than <span><math><mo>±</mo></math></span>1% across all signal-to-noise ratio levels and frequency ranges, collectively indicating enhanced accuracy and precision. Furthermore, SAGA-ST extends the usable bandwidth for attenuation estimation, compared to the 2D-FT-based method, offering improved accuracy for tissue characterization.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111042"},"PeriodicalIF":6.3,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118269","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":"CTBN-PH: A continuous-time Bayesian network for individualised diagnostic risk prediction","authors":"Guillem Hernández Guillamet ,&nbsp;Francesc López Seguí ,&nbsp;Josep Vidal Alaball ,&nbsp;Beatriz López","doi":"10.1016/j.compbiomed.2025.111069","DOIUrl":"10.1016/j.compbiomed.2025.111069","url":null,"abstract":"<div><div>Continuous-time Bayesian networks (CTBNs) are powerful tools for modelling and predicting complex disease trajectories in continuous-time scenarios. However, their application is often limited by a lack of individualisation in the results, if the covariates significantly influence a patient’s diagnostic transitions. To address these challenges, we introduce the CTBN-PH model, which integrates CTBN models with Cox proportional hazards (Cox-PH) models. The proposed model combines the dynamic and probabilistic capabilities of CTBNs with the robust, covariate-driven risk estimation of Cox-PH models. By leveraging causal topologies learned from healthcare trajectories, the method dynamically adjusts transition intensities based on covariate effects, enabling efficient parameter learning in extensive databases. We validated the model using a dataset of over 2.1 million patients and found that it learned complex causal structures associated with multi-morbid conditions such as diabetes and hypertension. Performance comparisons with non-individualised and non-causally inferred networks highlight the model’s effectiveness. Our model achieved an integrated Brier score (IBS) of 0.153 for predicting the onset of a single diagnosis over 25 years and an IBS of 0.04 for forecasting the inertia of the entire system over four years. Additionally, we explore the model’s utility in simulating patient trajectories that are tailored to specific covariate-defined populations.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111069"},"PeriodicalIF":6.3,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118157","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":"Energy concentration-based optimization of Gaussian window reassignment for time–frequency analysis, and its application in analyzing the non-stationary spectral and coherence of cardiovascular signals","authors":"Henok Mezemr Besfat ,&nbsp;Demissie Jobir Gelmecha ,&nbsp;Ram Sewak Singh","doi":"10.1016/j.compbiomed.2025.111040","DOIUrl":"10.1016/j.compbiomed.2025.111040","url":null,"abstract":"<div><div>The non-stationary nature of cardiovascular signals necessitates time–frequency (TF) methodologies that can precisely capture and characterize temporal variations in their frequency components. This study aims to explore the Non-Stationary Spectral Coherence (NSSC) characteristics between cardiovascular signals, including heart rate variability (HRV), respiratory (RESP), systolic arterial blood pressure variability (SABPV), and arterial pressure interval variability (APIV), of healthy young (yng) and elderly (eldy) subjects to understand cardiovascular control. For this analysis, an optimized Gaussian window reassignment (OptGWRS) TF technique is proposed to capture the characteristics of these signals. The parameters of the Gaussian window are optimized using a particle swarm optimization (PSO) algorithm to maximize the energy concentration of signals in the TF domain. Initially, the proposed method was validated using synthetic stationary and non-stationary time series data that flow the dynamic characteristics of cardiovascular signals within the low-frequency 0.04 to 0.15 Hz (LF) and high-frequency 0.15 to 0.4 Hz. (HF) band. The simulation results have demonstrated that the proposed method provided a more energy concentration and achieved a higher TF resolution. Furthermore, this method was applied to a real database. The simulation results have shown that the NSSC ranges between the HRV and APIV signals in the low frequency band has been significantly reduced (p = 0.000001) in the eldy group when compared to the yng group. In the LF band, the NSSC among HRV and RESP signals have not been influenced by aging and, in the HF band significantly reduced in the eldy compared to the yng group (p = 0.0125).</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111040"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118072","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":"Exploring transfer learning techniques for classifying Alzheimer's disease with rs-fMRI","authors":"Somayeh Abbasabadi ,&nbsp;Parviz Fattahi ,&nbsp;Mahdyeh Shiri","doi":"10.1016/j.compbiomed.2025.111075","DOIUrl":"10.1016/j.compbiomed.2025.111075","url":null,"abstract":"<div><div>Alzheimer's disease, the most prevalent form of dementia, leads to a fatal progression after progressively destroying memory at each stage. This irreversible disease appears more frequently in older populations. Even though research on Alzheimer's disease has risen over the past few years, the intricacy of brain structure and function creates challenges for accurate disease diagnosis. As a neuroimaging technology, resting-state functional magnetic resonance imaging enables researchers to study debilitating neural diseases while scanning the brain. The research investigates resting-state functional magnetic resonance imaging approaches and deep learning methods to distinguish between Alzheimer's patients and normal individuals. resting-state functional magnetic resonance imaging of 97 participants is obtained from the Alzheimer's disease neuroimaging initiative database, with 56 participants classified in the Alzheimer's disease group and 41 in the normal control group. Extensive preprocessing is applied to the resting-state functional magnetic resonance imaging data before classification. Using transfer learning, classification between the normal control and Alzheimer's disease groups is conducted with proposed VGG19, AlexNet, and ResNet50 algorithms; the classification accuracy of them is 96.91 %, 98.71 %, and 98.20 %, respectively. For evaluation, precision, recall, and F1-score are utilized as additional assessment metrics. The AlexNet model exhibits higher accuracy than the other models and outperforms them in other evaluation metrics, including precision, recall, and F1-score. While AlexNet achieves the highest overall classification performance, ResNet50 demonstrates superior interpretability through Grad-CAM visualizations, producing more anatomically focused and clinically meaningful attention maps.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111075"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118071","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":"High-throughput nucleotide sequencing reveals new circulatory miRNA genes in cervical cancer patients","authors":"Hussein Hameed Abbas,&nbsp;Fenwick Antony Edwin Rodrigues,&nbsp;Deena Krishnan,&nbsp;Sivasamy Ramasamy","doi":"10.1016/j.compbiomed.2025.111116","DOIUrl":"10.1016/j.compbiomed.2025.111116","url":null,"abstract":"<div><h3>Background</h3><div>Cervical cancer is one of the most common cancers in the world, and it is the second gynecological cancer in women after breast cancer. Recent evidence has shown miRNAs are dysregulated in cervical cancer, which can be used as a valuable tool for diagnosis, prognosis, or treatment. Aim: In this study, we aimed to find a new miRNA signature for cervical cancer and its potential target genes.</div></div><div><h3>Methods</h3><div>5.0 mL of whole blood were taken from 20 samples, 10 cervical cancer, and 10 healthy women, then we separated the serum and extracted RNA. The RNA samples of each group were pooled together. The differentially expressed miRNAs were found using high-throughput nucleotide sequencing. The target genes of the selected miRNAs were predicted by two tools (miRTarBase and TargetScan). The mRNA data of the target genes were downloaded from TCGA and processed with R version 4.4.1 for survival analysis.</div></div><div><h3>Results</h3><div>The study found two low-expressed miRNAs (miR-384 and miR-6880-5p) and three upregulated miRNAs (miR-325, miR-504-5p, and miR-3074-5p); those miRNAs are targeting 11 critical genes in cancer pathways. Survival analysis identified the target genes <em>CDK6</em> and <em>FGF19</em> as potential prognostic genes. The qRT-PCR results showed that the expression of the miRNAs miR-6880-5p and ch21_12598 was downregulated in the patient group.</div></div><div><h3>Conclusion</h3><div>The current study found five known miRNAs and one novel miRNA in cervical cancer that target essential genes in cancer pathways. These potential biomarkers can be further investigated for their potential role in cervical cancer diagnosis, prognosis, and treatment.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111116"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145112203","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":"Dynamic topographies of intrinsic neural timescales: a key role for consciousness","authors":"Andrea Buccellato ,&nbsp;Di Zang ,&nbsp;Yasir Çatal ,&nbsp;Bianca Ventura ,&nbsp;Massimiliano Facca ,&nbsp;Zengxin Qi ,&nbsp;Patrizia Bisiacchi ,&nbsp;Alessandra Del Felice ,&nbsp;Xuehai Wu ,&nbsp;Georg Northoff","doi":"10.1016/j.compbiomed.2025.111102","DOIUrl":"10.1016/j.compbiomed.2025.111102","url":null,"abstract":"<div><div>The brain displays intrinsic durations in its own spontaneous activity - Intrinsic Neural Timescales (INTs). INTs are hierarchically organized, with shorter durations within unimodal regions and longer intervals in multimodal domains. Despite significant progress, it's currently not known whether the unimodal-multimodal hierarchical organization undergoes recurrent changes itself - consistent with the existence of a dynamic repertoire of INT topographies. To this aim, we characterized the dynamics of topographic INT states by clustering the dynamic ACW-0 matrices in two different datasets: the source-reconstructed HCP resting-state MEG dataset, and a hd-EEG resting-state dataset, composed of healthy individuals and people with disorders of consciousness (DoCs). We found that healthy subjects display dynamic transitions between different INT states, which exhibit changing degrees of uni-transmodal cortical hierarchies. These dynamic transitions show non-random behavior, with moderate degrees of unpredictability and evidence of nontrivial memory effects. Unlike in healthy subjects, these properties are disrupted in DoC patients, who exhibit less predictable INT state transitions and less memory effects. Together, our results show a prominent role for the temporal richness of the transitions between different INT topographic states in the awake state which, as evidenced by our results, is key for maintaining an adequate level of consciousness.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111102"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145112225","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":"Strategic frameworks: A review of game theory methods for privacy preservation in digital health","authors":"Hamed Narimani ,&nbsp;Maryam Ansarian ,&nbsp;Zahra Baharlouei","doi":"10.1016/j.compbiomed.2025.111124","DOIUrl":"10.1016/j.compbiomed.2025.111124","url":null,"abstract":"<div><div>With the advancement of technology and the transition towards a digital world, the field of health and medicine is rapidly evolving in this direction. To fully harness the benefits of digital health, it is crucial to address the associated challenges. Given the necessity of exchanging personal information between patients and healthcare centers over communication networks, ensuring security and preserving privacy are important challenging issues in this field. Various approaches have been proposed in the literature to tackle these challenges. Some studies have utilized game theory to analyze and model the issues of security and privacy. Over recent decades, game theory has proven its versatility in modeling and solving a variety of problems. Research indicates that game theory can significantly enhance healthcare outcomes, having been utilized across numerous specialties such as disease diagnosis, public health, cancer treatment, medical consultations, clinical decision-making, privacy, and security of medical information. In this paper, for the first time, we review game-theory-based methods for preserving privacy and security in digital health, categorizing them based on the types of games modeled. Our study results show that the most commonly used game models in this field are, in order, the Stackelberg, the Strategic, and the Evolutionary games. Based on the research conducted in each category of games, we extract the common model used so that these models can be utilized in future research.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111124"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118160","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":"Developing federated time-to-event scores using heterogeneous real-world survival data","authors":"Siqi Li ,&nbsp;Ziwen Wang ,&nbsp;Yuqing Shang ,&nbsp;Qiming Wu ,&nbsp;Chuan Hong ,&nbsp;Yilin Ning ,&nbsp;Di Miao ,&nbsp;Marcus Eng Hock Ong ,&nbsp;Bibhas Chakraborty ,&nbsp;Nan Liu","doi":"10.1016/j.compbiomed.2025.111084","DOIUrl":"10.1016/j.compbiomed.2025.111084","url":null,"abstract":"<div><h3>Objective</h3><div>Survival analysis serves as a fundamental component in numerous healthcare applications, where the determination of the time to specific events (such as the onset of a certain disease or death) for patients is crucial for clinical decision-making. Scoring systems are widely used for swift and efficient risk prediction. However, existing methods for constructing survival scores presume that data originates from a single source, posing privacy challenges in collaborations with multiple data owners.</div></div><div><h3>Materials and methods</h3><div>We propose a novel framework for building federated scoring systems for multi-site survival outcomes, ensuring both privacy and communication efficiency. We applied our approach to sites with heterogeneous survival data originating from emergency departments in Singapore and the United States. Additionally, we independently developed local scores at each site.</div></div><div><h3>Results</h3><div>In testing datasets from each participant site, our proposed federated scoring system consistently outperformed all local models, evidenced by higher integrated area under the receiver operating characteristic curve (iAUC) values, with a maximum improvement of 11.6 %. Additionally, the federated score's time-dependent AUC(t) values showed advantages over local scores, exhibiting narrower confidence intervals (CIs) across most time points.</div></div><div><h3>Discussion</h3><div>The model developed through our proposed method showed good local performance and is promising for future healthcare research. Sites participating in our proposed federated scoring model training can develop survival models with enhanced prediction accuracy and efficiency.</div></div><div><h3>Conclusion</h3><div>This study demonstrates the effectiveness of our privacy-preserving federated survival score generation framework and its applicability to real-world heterogeneous survival data.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111084"},"PeriodicalIF":6.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145112195","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}