Computers in biology and medicine最新文献

{"title":"Does maternal respiration modulate maternal-fetal cardiovascular coupling?","authors":"G Steyde, A Galli, E Peri, M B Van der Hout-van der Jagt, J O E H Van Laar, M G Signorini, M Mischi","doi":"10.1016/j.compbiomed.2026.111627","DOIUrl":"https://doi.org/10.1016/j.compbiomed.2026.111627","url":null,"abstract":"<p><strong>Objective: </strong>Maternal Fetal Cardiac Coupling (MFCC) is an understudied physiological phenomenon that could play a key role in the understanding and early detection of pregnancy-related pathologies. In this study, we aimed to investigate the role of maternal respiration in the short-term directed interactions between the maternal heart rate (MHR) and fetal heart rate (FHR) in healthy fetuses. Additionally, we investigated how such coupling patterns might be impaired in fetuses affected by conotruncal anomalies (CTA).</p><p><strong>Methods: </strong>FHR, MHR and maternal respiration were extracted from non-invasive abdominal electrophysiological recordings from over 300 pregnant women in the second trimester of gestation. Linear coupling metrics-Directed Coherence (DC), Partial Directed Coherence (PDC), and generalized PDC (gPDC)-were computed with and without including the respiratory signal in the Multivariate Autoregressive model. Results between healthy and CTA fetuses were compared.</p><p><strong>Results: </strong>Maternal respiration Granger-causes both the FHR and the MHR, suggesting respiration is a dominant source of directional influence in MFCC during the second trimester. Preliminary results suggest that the directed interaction from maternal respiration to the FHR is reduced in fetuses with CTA.</p><p><strong>Conclusion: </strong>Maternal respiration plays a main role in the physiological phenomenon of MFCC.</p><p><strong>Significance: </strong>This work sheds light on a fundamental dynamic regulating MFCC, which might improve the monitoring of pregnancy-related cardiovascular pathologies and pave the way for their early diagnosis.</p>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"207 ","pages":"111627"},"PeriodicalIF":6.3,"publicationDate":"2026-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147485089","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":"Ollama-driven medical insights using LLMs with a federated learning approach","authors":"Gurbaksh Lal ,&nbsp;Geetanjali Rathee ,&nbsp;Chaker Abdelaziz Kerrache","doi":"10.1016/j.compbiomed.2026.111514","DOIUrl":"10.1016/j.compbiomed.2026.111514","url":null,"abstract":"<div><div>Traditional medical diagnostics often suffer from delays and inconsistencies due to the manual interpretation of unstructured patient data. To overcome these challenges, we introduce Our Model (given name as ‘AI Doctor’)—a novel diagnostic system built on the Ollama platform that integrates multiple pretrained large language models (Meditron, MedLLaMA2, WizardLM2, and Mistral) through an innovative prompt filtering mechanism. AI Doctor accurately interprets patient-reported symptoms to deliver precise diagnoses and personalized treatment recommendations, while its design supports robust local deployment and includes a theoretical framework for federated learning. This federated approach facilitates decentralized, privacy-preserving model updates across healthcare institutions. Performance evaluations using BLEU scores, structured output analysis, and inference speed measurements demonstrate that AI Doctor consistently outperforms individual models, ensuring high diagnostic accuracy and realtime clinical applicability.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111514"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076948","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":"Periodontal bone loss analysis via keypoint detection with heuristic post-processing","authors":"Ryan Banks ,&nbsp;Vishal Thengane ,&nbsp;María Eugenia Guerrero ,&nbsp;Nelly Maria García-Madueño ,&nbsp;Yunpeng Li ,&nbsp;Hongying Tang ,&nbsp;Akhilanand Chaurasia","doi":"10.1016/j.compbiomed.2026.111515","DOIUrl":"10.1016/j.compbiomed.2026.111515","url":null,"abstract":"<div><div><em><strong>Objectives:</strong></em> This study proposes a deep learning framework and an annotation methodology for the automatic detection of periodontal bone loss landmarks, associated conditions, and staging.</div><div><em><strong>Methods</strong></em> <span><math><mn>192</mn></math></span> periapical radiographs were collected and annotated using a stage agnostic methodology, labelling clinically relevant landmarks regardless of disease presence or extent. We propose a heuristic post-processing module that aligns predicted keypoints to tooth boundaries using an auxiliary instance segmentation model. An evaluation metric, Percentage of Relative Correct Keypoints (<span><math><mi>P</mi><mi>R</mi><mi>C</mi><mi>K</mi></math></span>), is proposed to capture keypoint performance in dental imaging domains. Four donor pose estimation models were adapted with fine-tuning for our keypoint problem.</div><div><em><strong>Results</strong></em> Post-processing improved fine-grained localisation, raising average <span><math><mi>P</mi><mi>R</mi><mi>C</mi><msup><mi>K</mi><mrow><mn>0.05</mn></mrow></msup></math></span> by <span><math><mo>+</mo><mn>0.028</mn></math></span>, but reduced coarse performance for <span><math><mi>P</mi><mi>R</mi><mi>C</mi><msup><mi>K</mi><mrow><mn>0.25</mn></mrow></msup></math></span> by <span><math><mo>−</mo><mn>0.0523</mn></math></span> and <span><math><mi>P</mi><mi>R</mi><mi>C</mi><msup><mi>K</mi><mrow><mn>0.5</mn></mrow></msup></math></span> by <span><math><mo>−</mo><mn>0.0345</mn></math></span>. Orientation estimation shows excellent performance for auxiliary segmentation when filtered with either stage 1 object detection model. Periodontal staging was detected sufficiently, with the best mesial and distal Dice scores of <span><math><mn>0.508</mn></math></span> and <span><math><mn>0.489</mn></math></span>, while furcation involvement and widened periodontal ligament space tasks remained challenging due to scarce positive samples. Scalability is implied with similar validation and external set performance.</div><div><em><strong>Conclusion</strong></em> The annotation methodology enables stage agnostic training with balanced representation across disease severities for some detection tasks. The <span><math><mi>P</mi><mi>R</mi><mi>C</mi><mi>K</mi></math></span> metric provides a domain-specific alternative to generic pose metrics, while the heuristic post-processing module consistently corrected implausible predictions with occasional catastrophic failures.</div><div><em><strong>Clinical significance:</strong></em> The proposed framework demonstrates the feasibility of clinically interpretable periodontal bone loss assessment, with the potential to reduce diagnostic variability and clinician workload.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111515"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146156472","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":"IUPAC-induced computational approaches for identifying boosters of small biomolecule functionality: A case study of human tyrosyl-DNA phosphodiesterase 1 (TDP1) inhibitors","authors":"Mariya L. Ivanova ,&nbsp;Nicola Russo ,&nbsp;Gueorgui Mihaylov ,&nbsp;Konstantin Nikolic","doi":"10.1016/j.compbiomed.2026.111531","DOIUrl":"10.1016/j.compbiomed.2026.111531","url":null,"abstract":"<div><div>This paper introduces several proof-of-concept (PoC) computational methods intended to offer biochemical researchers straightforward, time- and cost-effective strategies to accelerate their work. While Machine Learning (ML) models were developed, the study's central purpose was to explore approaches for the identification of desirable functional groups/fragments in small biomolecules regarding a specific functionality, which, in this case, was human tyrosyl-DNA phosphodiesterase 1 (TDP1) inhibition. This was achieved primarily by tokenising IUPAC names to generate features. Additionally, the applicability of the CID_SID ML model for predicting TDP1 activity was developed and explored. Since these computational approaches were not experimentally validated due to a lack of appropriate laboratory facilities, they are presented as open proposals for further laboratory investigation.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111531"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146130605","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":"Clinical pathways discovery for long-term and chronic patients: A process mining approach","authors":"Luca Murazzano ,&nbsp;Paolo Landa ,&nbsp;Jean–Baptiste Gartner ,&nbsp;Mohamed Hakim Raki ,&nbsp;André Côté","doi":"10.1016/j.compbiomed.2026.111539","DOIUrl":"10.1016/j.compbiomed.2026.111539","url":null,"abstract":"<div><div>The increasing burden of chronic respiratory diseases is placing substantial pressure on healthcare systems around the world. Lung diseases, such as chronic obstructive pulmonary disease, pneumonia, lung cancer, and pulmonary fibrosis, rank among the most prevalent and deadly conditions globally. Given the complexity of managing these chronic conditions, there is an urgent need to optimize care processes to meet the growing service demands efficiently and effectively, especially in public healthcare systems where there is a prevalence of elderly patients. This study aims to understand and improve the quality and performance of treatments provided to patients using a process mining approach. By analyzing clinical data collected from 2018 to 2022, the study identifies critical points in the care pathways where improvements can be made. This approach enables the optimization of resource deployment and service configuration to better meet patient needs. As a case study, this method was applied to a specialized hospital facility dedicated to cardiac and respiratory diseases, where actionable insights were uncovered for enhancing clinical pathways. This study allows us to analyze clinical pathways and detect critical points, providing insights to healthcare managers and decision makers. In addition, it highlights the importance of adequate data collection and suggests that future research efforts should prioritize the acquisition of larger and more diverse data sets to enhance the reliability and validity of activity and episodes analyses.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111539"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146218811","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":"Corrigendum to “Regional attention-enhanced vision transformer for accurate Alzheimer's disease classification using sMRI data” [Comput. Biol. Med. 197 (2025) 111065]","authors":"Alireza Jomeiri ,&nbsp;Ahmad Habibizad Navin ,&nbsp;Mahboubeh Shamsi","doi":"10.1016/j.compbiomed.2026.111559","DOIUrl":"10.1016/j.compbiomed.2026.111559","url":null,"abstract":"","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111559"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146218872","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":"Unobtrusive sleep posture estimation using pressure sensor in home sleep","authors":"Jonghyun Hong ,&nbsp;Jungmin Koh ,&nbsp;Jinyoung Kim ,&nbsp;Hyunchan Ryu ,&nbsp;Dahye Lee ,&nbsp;Hyun Bin Kwon ,&nbsp;Byunghun Choi ,&nbsp;Heesu Park ,&nbsp;Kwang Suk Park ,&nbsp;Heenam Yoon","doi":"10.1016/j.compbiomed.2026.111551","DOIUrl":"10.1016/j.compbiomed.2026.111551","url":null,"abstract":"<div><h3>Purpose</h3><div>Sleep posture is associated with various physiological indicators and significantly influences sleep health and quality. Although several methods for posture estimation have been proposed, most have been evaluated using data from controlled laboratory environments. This study proposes a method for determining sleep posture in real-world settings using pressure sensor data.</div></div><div><h3>Methods</h3><div>The approach was developed based on data collected from 22 participants in a laboratory setting using a 7 × 14 array of force-sensitive resistors (FSR). We employed a support vector machine to classify four sleep postures—supine, left-lateral, right-lateral, and prone—based on six extracted features related to area, curvature, and row length ratio. The algorithm was subsequently evaluated using FSR data recorded from ten participants sleeping freely in their home environments.</div></div><div><h3>Results</h3><div>The performance results demonstrated an accuracy of 78.1% and a Cohen's kappa of 0.71 for the laboratory data. When applied to the home-environment data, the method achieved an accuracy of 86.1% and a Cohen's kappa of 0.76 for the classification of the four sleep postures.</div></div><div><h3>Conclusion</h3><div>These findings indicate that the model trained in a laboratory setting maintained high performance in real-world conditions, supporting the feasibility of implementing sleep monitoring technologies in daily life and clinical contexts. This study contributes to the development of noninvasive, long-term sleep monitoring systems and highlights the potential for future clinical applications in embedded systems and hospital environments through the use of feature-based models with high explainability.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111551"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171750","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":"External validation of GDM risk prediction models using a machine learning reciprocal model-exchange framework","authors":"Mark Germaine ,&nbsp;Yitayeh Belsti ,&nbsp;Amy O'Higgins ,&nbsp;Brendan Egan ,&nbsp;Helena Teede ,&nbsp;Graham Healy ,&nbsp;Joanne Enticott","doi":"10.1016/j.compbiomed.2026.111547","DOIUrl":"10.1016/j.compbiomed.2026.111547","url":null,"abstract":"<div><h3>Background</h3><div>Although many risk prediction models have been developed, very few undergo external validation, primarily due to issues with data access. Therefore, we implemented a reciprocal model-exchange approach to facilitate external validation and demonstrate its use with gestational diabetes mellitus (GDM) prediction models.</div></div><div><h3>Objective</h3><div>To assess the robustness and generalisability of two independently developed GDM risk prediction models using a reciprocal model-exchange framework.</div></div><div><h3>Methods</h3><div>Two independently developed GDM risk prediction models were externally validated using a reciprocal model-exchange. The saved model's corresponding variable types and data pre-processor were exchanged. The Monash CatBoost model was validated using Irish data at Dublin City University (DCU), and the DCU logistic-regression GDM model was validated using Australian data at Monash University. Performance was assessed using discrimination, calibration and decision curve analysis. Model fairness was assessed.</div></div><div><h3>Results</h3><div>The prevalence of GDM was 21.1% in the Australian cohort and 11.7% in the Irish cohort. The Monash model's AUC dropped from 0.93 to 0.77, while the DCU model's AUC fell from 0.82 to 0.69. Calibration estimates confirmed systematic risk misestimation; each model tends to over or under-predict GDM probabilities outside its training domain, with calibration-in-the-large of −0.573 for the Monash model and 0.17 for the DCU model; slopes were 1.278 and 0.55 respectively. Both models showed performance variability across ethnic groups, with lower performance for Southeast/Northeast Asians and both performed better with increasing parity and among women without a prior GDM diagnosis.</div></div><div><h3>Conclusions</h3><div>Each model's performance decreased upon external validation, and the fairness evaluations on the different sub-categories (ethnicities; parity and previous GDM) provided evidence on the areas to be addressed in model recalibration/updating before deployment can be progressed. This reciprocal model-exchange approach provides a solution to facilitating external validations, which are notably lacking in the current literature but are necessary to advance the risk prediction field.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111547"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146171747","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":"Analysis of EEG univariate features for epileptic seizures","authors":"Sergio E. Sánchez-Hernández ,&nbsp;Ricardo A. Salido-Ruiz ,&nbsp;Stewart R. Santos-Arce ,&nbsp;Radu Ranta","doi":"10.1016/j.compbiomed.2026.111517","DOIUrl":"10.1016/j.compbiomed.2026.111517","url":null,"abstract":"<div><div>The diagnosis of epilepsy and the identification of seizures are subject to multiple challenges. Hence, it is relevant to identify EEG features that allow for differentiation between seizure intervals and between patients and healthy subjects. Several studies have explored the search for biomarkers by utilizing signal processing techniques, although some of these studies have been applied to limited datasets. In this study, a set of seven univariate features in the time and frequency domains was calculated for scalp EEG recordings. These recordings were collected from four EEG datasets (patients=180, controls=100, seizures=613), and two types of experiments were performed: a comparison between healthy subjects and the pre-ictal (one min before seizure onset) and ictal intervals, and a comparison between seizure stages (pre-ictal, ictal and post-ictal). Variations in the normalized power spectral density were the most reliable indicator of seizure activity (<span><math><mi>p</mi><mo>−</mo><mi>v</mi><mi>a</mi><mi>l</mi><mi>u</mi><mi>e</mi><mo>&lt;</mo><mn>0.05</mn></math></span>). Mobility, complexity, and approximate entropy also changed significantly, with entropy-based measurements decreasing during seizures, indicating a reduction in EEG irregularity (<span><math><mi>p</mi><mo>−</mo><mi>v</mi><mi>a</mi><mi>l</mi><mi>u</mi><mi>e</mi><mo>&lt;</mo><mn>0.05</mn></math></span>). The results highlighted the importance of combining spectral, statistical, and entropy-based features for a more comprehensive understanding of seizures. Although some common patterns were identified, distinct behaviors were observed between datasets. Future work will benefit from a diverse and curated dataset; therefore, causes of dissimilarities can be unequivocally identified.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111517"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146112060","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":"A semi-automated modelling pipeline to predict the mechanics of multiple sclerosis lesion afflicted brains from magnetic resonance images","authors":"Adam C Szekely-Kohn ,&nbsp;Diana Cruz De Oliveira ,&nbsp;Marco Castellani ,&nbsp;Michael Douglas ,&nbsp;Zubair Ahmed ,&nbsp;Daniel M Espino","doi":"10.1016/j.compbiomed.2026.111519","DOIUrl":"10.1016/j.compbiomed.2026.111519","url":null,"abstract":"<div><div>Multiple Sclerosis (MS) is a demyelinating and degenerative autoimmune disease that affects the brain and spinal cord. Its causes, mechanisms, and outcomes are yet to be fully understood. One relatively unexplored area is the understanding of changes in brain biomechanics during MS disease progression, despite the likelihood that demyelination significantly alters the overall mechanical structure of the brain. Such changes have the potential to hinder the propagation of nerve signals essential for cognition and motor function. The aim of this work was to create a computational model to explore the mechanics of brains with MS, separating the brain into grey matter, white matter and lesions. Changes were observed when the surface of the brain was subjected to a ramped uniform pressure tangential to the faces of a finite element model, generated from patient- and time-specific MRI scans. The resulting displacements, stresses and strains can all be gauged using the model. The key benefit of this study was to observe the impact of changes in tissue morphology in real brains using non-invasive methods. Ensuring the accuracy of the axiomatic input tissue parameters of the models was critically important, as exploring the range of values from literature, adjusted by their error margins, revealed a significant variability in outcomes, especially in the case of volumetric strain of lesions. The model has the potential to track changes in mechanical tissue properties assuming the availability of a longitudinal dataset, and if further developed, has the potential to serve as the foundation for creating a digital twin. This could enhance medical practice and provide a non-invasive approach to advancing the understanding of MS and its progression on a patient-specific basis.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"204 ","pages":"Article 111519"},"PeriodicalIF":6.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146124039","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}