Computers in biology and medicine最新文献

{"title":"MHS U-Net: Multi-scale hybrid subtraction network for medical image segmentation","authors":"Junran Qian ,&nbsp;Haiyan Li ,&nbsp;Shuran Liao ,&nbsp;Zhe Xiao ,&nbsp;Weihua Li ,&nbsp;Hongsong Li","doi":"10.1016/j.compbiomed.2025.110431","DOIUrl":"10.1016/j.compbiomed.2025.110431","url":null,"abstract":"<div><div>Medical image segmentation plays a critical role in modern clinical diagnosis. However, existing methods face challenges such as insufficient feature extraction, limited spatial modeling capabilities, and restricted generalization ability with low computational cost. To address these challenges, we propose a Multi-scale Hybrid Subtraction Network (MHS U-Net) for Medical Image Segmentation. First, a pretrained PVTv2-B2 is integrated as the encoder to enhance the model's adaptability and feature extraction capability for complex multi-modal medical images. Second, a Multi-Layer Shift Perception Attention (MSPA) mechanism is designed at the bottleneck to capture fine-grained high-level features by deepening the network structure, while effectively suppressing the surge in computational cost through shift operations. In the decoder, a Multi-Scale Hybrid Convolution Subtraction Decoder (MSHCSD) is developed, to improve the modeling of spatial relationships within and around lesions and significantly enhance the model's generalization ability through integrating group convolution, gating mechanisms, and deep convolutional blocks. Additionally, to address the insufficient utilization of multi-scale feature interactions, a Multi-Scale Subtraction Module (MSSM) is proposed to strengthen cross-scale feature fusion through differential information extraction and complementary feature enhancement, thereby achieving the precise localization and segmentation of lesion regions. Experimental results on 14 public datasets across five imaging modalities demonstrate that MHS U-Net consistently outperforms state-of-the-art methods in metrics and visual results. Moreover, MHS U-Net requires only 5.48G FLOPs and 11.59M parameters, significantly lower than most existing models. Overall, MHS U-Net offers an excellent balance between model performance and size in multi-modal medical image segmentation tasks.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110431"},"PeriodicalIF":7.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144138782","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":"Advancing emotion recognition with Virtual Reality: A multimodal approach using physiological signals and machine learning","authors":"Edoardo Maria Polo ,&nbsp;Francesco Iacomi ,&nbsp;Alberto Valdes Rey ,&nbsp;Davide Ferraris ,&nbsp;Alessia Paglialonga ,&nbsp;Riccardo Barbieri","doi":"10.1016/j.compbiomed.2025.110310","DOIUrl":"10.1016/j.compbiomed.2025.110310","url":null,"abstract":"<div><h3>Introduction</h3><div>: Emotion recognition systems have traditionally relied on basic visual elicitation. Virtual reality (VR) offers an immersive alternative that better resembles real-world emotional experiences.</div></div><div><h3>Objective:</h3><div>To develop and evaluate custom-built VR scenarios designed to evoke sadness, relaxation, happiness, and fear, and to utilize physiological signals together with machine learning techniques for accurate prediction and classification of emotional states.</div></div><div><h3>Methods:</h3><div>Physiological signals (electrocardiogram, blood volume pulse, galvanic skin response, and respiration) were acquired from 36 participants during VR experiences. Machine learning models, including Logistic Regression with Square Method feature selection, were applied in a subject-independent approach in order to discern the four emotional states.</div></div><div><h3>Results:</h3><div>Features extracted by physiological signal analysis highlighted significant differences among emotional states. The machine learning models achieved high accuracies of 80%, 85%, and 70% for arousal, valence, and 4-class emotion classification, respectively. Explainable AI techniques further provided insights into the decision-making processes and the relevance of specific physiological features, with galvanic skin response peaks emerging as the most significant feature for both valence and arousal dimensions.</div></div><div><h3>Conclusion:</h3><div>The proposed study demonstrates efficacy of VR in eliciting genuine emotions and the potential of using physiological signals for emotion recognition, with important implications for affective computing and psychological research. The non-invasive approach, robust subject-independent generalizability, and compatibility with wearable technology position this methodology favorably for practical applications in mental health contexts and user experience evaluation.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110310"},"PeriodicalIF":7.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144138776","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":"Brain Fractal Dimension and Machine Learning can predict first-episode psychosis and risk for transition to psychosis","authors":"Yaxin Hu ,&nbsp;Marina Frisman ,&nbsp;Christina Andreou ,&nbsp;Mihai Avram ,&nbsp;Anita Riecher-Rössler ,&nbsp;Stefan Borgwardt ,&nbsp;Erhardt Barth ,&nbsp;Alexandra Korda","doi":"10.1016/j.compbiomed.2025.110333","DOIUrl":"10.1016/j.compbiomed.2025.110333","url":null,"abstract":"<div><div>Although there are notable structural abnormalities in the brain associated with psychotic diseases, it is still unclear how these abnormalities relate to clinical presentation. However, the fractal dimension (FD), which offers details on the complexity and irregularity of brain microstructures, may be a promising feature, as demonstrated by neuropsychiatric disorders such as Parkinson’s and Alzheimer’s. It may offer a possible biomarker for the detection and prognosis of psychosis when paired with machine learning. The purpose of this study is to investigate FD as a structural magnetic resonance imaging (sMRI) feature from individuals with a high clinical risk of psychosis who did not transit to psychosis (CHR_NT), clinical high risk who transit to psychosis (CHR_T), patients with first-episode psychosis (FEP) and healthy controls (HC). Using a machine learning approach that ultimately classifies sMRI images, the goals are (a) to evaluate FD as a potential biomarker and (b) to investigate its ability to predict a subsequent transition to psychosis from the high-risk clinical condition. We obtained sMRI images from 194 subjects, including 44 HCs, 77 FEPs, 16 CHR_Ts, and 57 CHR_NTs. We extracted the FD features and analyzed them using machine learning methods under five classification schemas (a) FEP vs. HC, (b) FEP vs. CHR_NT, (c) FEP vs. CHR_T, (d) CHR_NT vs. CHR_T, (d) CHR_NT vs. HC and (e) CHR_T vs. HC. In addition, the CHR_T group was used as external validation in (a), (b) and (d) comparisons to examine whether the progression of the disorder followed the FEP or CHR_NT patterns. The proposed algorithm resulted in a balanced accuracy greater than 0.77. This study has shown that FD can function as a predictive neuroimaging marker, providing fresh information on the microstructural alterations triggered throughout the course of psychosis. The effectiveness of FD in the detection of psychosis and transition to psychosis should be established by further research using larger datasets.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110333"},"PeriodicalIF":7.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134025","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":"Beyond Accuracy: Evaluating certainty of AI models for brain tumour detection","authors":"Zaib Un Nisa ,&nbsp;Sohail Masood Bhatti ,&nbsp;Arfan Jaffar ,&nbsp;Tehseen Mazhar ,&nbsp;Tariq Shahzad ,&nbsp;Yazeed Yasin Ghadi ,&nbsp;Ahmad Almogren ,&nbsp;Habib Hamam","doi":"10.1016/j.compbiomed.2025.110375","DOIUrl":"10.1016/j.compbiomed.2025.110375","url":null,"abstract":"<div><div>Brain tumors pose a severe health risk, often leading to fatal outcomes if not detected early. While most studies focus on improving classification accuracy, this research emphasizes prediction certainty, quantified through loss values. Traditional metrics like accuracy and precision do not capture confidence in predictions, which is critical for medical applications. This study establishes a correlation between lower loss values and higher prediction certainty, ensuring more reliable tumor classification.</div><div>We evaluate CNN, ResNet50, XceptionNet, and a Proposed Model (VGG19 with customized classification layers) using accuracy, precision, recall, and loss. Results show that while accuracy remains comparable across models, the Proposed Model achieves the best performance (96.95 % accuracy, 0.087 loss), outperforming others in both precision and recall. These findings demonstrate that certainty-aware AI models are essential for reliable clinical decision-making.</div><div>This study highlights the potential of AI to bridge the shortage of medical professionals by integrating reliable diagnostic tools in healthcare. AI-powered systems can enhance early detection and improve patient outcomes, reinforcing the need for certainty-driven AI adoption in medical imaging.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110375"},"PeriodicalIF":7.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144138777","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":"Integrating AI/ML and multi-omics approaches to investigate the role of TNFRSF10A/TRAILR1 and its potential targets in pancreatic cancer","authors":"Sudhanshu Sharma ,&nbsp;Rajesh Singh ,&nbsp;Shiva Kant ,&nbsp;Manoj K. Mishra","doi":"10.1016/j.compbiomed.2025.110432","DOIUrl":"10.1016/j.compbiomed.2025.110432","url":null,"abstract":"<div><div>Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, with a five-year survival of under 10 % despite current therapies. Aggressive tumor biology, a desmoplastic stroma that limits drug delivery and immune cell infiltration, and profound resistance to apoptosis make it more complex to treat. Here, we describe a multi-layered system biology and drug discovery pipeline that integrates bulk genomics, single-cell spatial transcriptomics, proteomics, competing endogenous RNA (ceRNA) network analysis, and deep learning-driven quantitative structure-activity relationship (QSAR) modeling. By implementing this pipeline, we predicted that TNFRSF10A encodes for the TRAILR1 death receptor as a potential therapeutic target in PDAC. Mutational and expressional analysis also confirmed TNFRSF10A as a putative target in PDAC. Cancer cells within the PDAC microenvironment exhibit aberrantly elevated TNFRSF10A expression. Immune-excluded tumor niches and pro-survival signaling link this elevated expression. Using an advanced transformer-based deep learning approach, SELFormer, combined with QSAR analysis-based virtual screening, we identified previously unexplored FDA-approved drugs and natural compounds, i.e., Temsirolimus, Ergotamine, and capivasertib, with potential TRAILR1 modulatory effects. During molecular dynamics simulations, these repurposed candidates showed the highest binding affinities against TNFRSF10A for 300 ns. These showed favorable binding energies (MM-PBSA), minimal RMSD drift, PCA, and SASA. We propose TNFRSF10A as a therapeutically important PDAC vulnerability nurtured by spatially resolved expression patterns and dynamic molecular modeling. This study has used a novel integration of AI-implemented chemical modeling, high-throughput screening, and a multi-omics approach to unravel and pharmacologically target a cancer compartment-specific weakness in a notoriously drug-resistant cancer.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110432"},"PeriodicalIF":7.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134024","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":"COPS4 is a novel prognostic biomarker and potential therapeutic target involved in regulation of immune microenvironment in numerous cancers","authors":"Abdul Jamil Khan ,&nbsp;Shahid Ullah Khan","doi":"10.1016/j.compbiomed.2025.110400","DOIUrl":"10.1016/j.compbiomed.2025.110400","url":null,"abstract":"<div><div>CHOL, HNSC, ESCA, and LIHC are among the most aggressive and fatal malignancies worldwide. Despite their clinical burden, these cancers still lack dependable biomarkers for early detection, prognosis, and therapeutic targeting. The COP9 signalosome (COPS), a key regulator of the ubiquitin proteasome pathway, has been shown to be aberrantly expressed in various cancer types and is thought to contribute to tumor development and progression. Among its subunits, COPS4 plays an essential role in maintaining the functional integrity of the complex. However, its prognostic significance and clinicopathological relevance in cancer remain poorly understood.</div><div>This study adopted a comprehensive, integrative bioinformatics framework grounded in TCGA-derived datasets, incorporating analytical platforms encompassing UALCAN, GEPIA, MEXPRESS, OncoDB, UCSC Xena, ENCORI, TIMER, GeneMANIA, TNMplot, and TISIDB. Through this strategy, the investigation delineated the transcriptional landscape, genomic aberrations, immunological associations, and putative functional roles of COPS4 across CHOL, HNSC, ESCA, and LIHC. Virtual screening and molecular dynamics simulations were performed to explore its druggable potential. COPS4 expression was significantly upregulated in tumor tissues and exhibited strong associations with key clinical parameters, including pathological stage, histological grade, nodal involvement, and metastatic status. IHC analysis further validated elevated protein levels in tumor specimens compared to adjacent non-neoplastic tissues. Genomic alterations were frequent, with predominant mutations in LIHC, amplifications in CHOL, and both amplifications and deletions in HNSC and ESCA. COPS4 expression showed positive associations with subset of oncogenes and inverse correlations with tumor suppressors. Notably, NUP54 and HELQ emerged as consistent co-targets. Immune analysis revealed strong positive correlations between COPS4 and infiltrating immune cells, including CD8⁺ and CD4⁺ T cells, B cells, macrophages, neutrophils, and dendritic cells. Somatic copy number variations of COPS4 also influenced immune cell infiltration and patient survival outcomes. Promoter hypomethylation and gene amplification were identified as mechanisms driving its overexpression. Finally, virtual screening and molecular dynamics simulations identified FDA-approved drugs interacting with COPS4, emphasizing its oncogenic role and therapeutic potential.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110400"},"PeriodicalIF":7.0,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144138779","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":"Speech signals-based Parkinson’s disease diagnosis using hybrid autoencoder-LSTM models","authors":"Ayşe Nur Tekindor ,&nbsp;Eda Akman Aydın","doi":"10.1016/j.compbiomed.2025.110334","DOIUrl":"10.1016/j.compbiomed.2025.110334","url":null,"abstract":"<div><div>Parkinson’s disease (PD) is a neurodegenerative disorder that occurs as a result of a decrease in the chemical called dopamine in the brain. There is no definitive treatment for PD, but some medications used to control symptoms in the early stages have a critical effect on the progression of the disease. Approximately 90% of patients with PD have vocal problems, and although voice disorders seen in the early stages are not apparent in the patient’s speech, they can be detected by acoustic analysis. In this study, a decision support system was proposed for the diagnosis of PD utilizing the feature extraction power of autoencoder (AE) &amp; long short-term memory (LSTM) models by using speech signals as input data. Firstly, simple (SAE), convolutional (CAE), and recurrent (RAE) AE models were created for the ablation analysis. Then, the effect of hybridization and deepening of these models with LSTM layers on the classification performance was observed. Within the scope of the study, RAE achieved the highest accuracy among the base models while CAE &amp; LSTM hybrid model provided the highest performance among all models with 95.79% accuracy for PD diagnosis based on audio signals. It was concluded that hybridization of the AE and LSTM models significantly improved the performance of simple and convolutional AE, and deepening the network to a certain extent improves the classification performance according to the type of AE.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110334"},"PeriodicalIF":7.0,"publicationDate":"2025-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144130984","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":"Relational Bi-level aggregation graph convolutional network with dynamic graph learning and puzzle optimization for Alzheimer's classification","authors":"K. Raajasree,&nbsp;R. Jaichandran","doi":"10.1016/j.compbiomed.2025.110292","DOIUrl":"10.1016/j.compbiomed.2025.110292","url":null,"abstract":"<div><div>Alzheimer's disease (AD) is a neurodegenerative disorder characterized by a progressive cognitive decline, necessitating early diagnosis for effective treatment. This study presents the <strong>Relational Bi-level Aggregation Graph Convolutional Network with Dynamic Graph Learning and Puzzle Optimization for Alzheimer's Classification (RBAGCN-DGL-PO-AC), using denoised T1-weighted Magnetic Resonance Images (MRIs)</strong> collected from Alzheimer's Disease Neuroimaging Initiative (ADNI) repository. Addressing the impact of noise in medical imaging, the method employs advanced denoising techniques includes: the <strong>Modified Spline-Kernelled Chirplet Transform (MSKCT), Jump Gain Integral Recurrent Neural Network (JGIRNN),</strong> and <strong>Newton Time Extracting Wavelet Transform (NTEWT),</strong> to enhance the image quality. Key brain regions, crucial for classification such as hippocampal, lateral ventricle and posterior cingulate cortex are segmented using Attention Guided <strong>Generalized Intuitionistic Fuzzy C-Means Clustering (</strong>AG-GIFCMC<strong>)</strong>. Feature extraction and classification using segmented outputs are performed with RBAGCN-DGL and puzzle optimization, categorize input images into Healthy Controls (HC), Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI), and Alzheimer's Disease (AD). To assess the effectiveness of the proposed method, we systematically examined the structural modifications to the RBAGCN-DGL-PO-AC model through extensive ablation studies. Experimental findings highlight that RBAGCN-DGL-PO-AC state-of-the art performance, with 99.25 % accuracy, outperforming existing methods including MSFFGCN_ADC, CNN_CAD_DBMRI, and FCNN_ADC, while reducing training time by 28.5 % and increasing inference speed by 32.7 %. Hence, the RBAGCN-DGL-PO-AC method enhances AD classification by integrating denoising, segmentation, and dynamic graph-based feature extraction, achieving superior accuracy and making it a valuable tool for clinical applications, ultimately improving patient outcomes and disease management.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110292"},"PeriodicalIF":7.0,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124319","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":"Weakly-supervised semantic segmentation in histology images using contrastive learning and self-training","authors":"Reza Safdari ,&nbsp;Dariush Lotfi ,&nbsp;Mahmood Amiri ,&nbsp;Herbert Peremans","doi":"10.1016/j.compbiomed.2025.110321","DOIUrl":"10.1016/j.compbiomed.2025.110321","url":null,"abstract":"<div><div>This paper presents a novel method for weakly-supervised semantic segmentation (WSSS) of histology images, where only global image-level labels are employed. We leverage an existing weakly-supervised object localization (WSOL) method to generate class activation maps (CAMs) indicating the spatial locations of relevant tissue regions. Next, we utilize a specialized encoder-decoder network to predict fine localization masks. A pixel-wise contrastive loss function is introduced to encourage the model to learn discriminative features for foreground and background regions. Additionally, a pixel-wise cross-entropy loss is incorporated for improved pixel-level supervision. An offline multi-round self-training strategy is also proposed to iteratively refine pseudo masks, enhancing segmentation performance. Our method demonstrates superior segmentation accuracy over the state-of-the-art method on the GlaS dataset (public benchmark for colon cancer). Furthermore, we investigate the efficacy of our approach in a mixed-supervision setting, achieving performance comparable to fully supervised models, indicating its practical applicability in clinical settings. Our results show that the proposed method offers an effective and practical solution for weakly-supervised semantic segmentation in histology images, potentially aiding pathologists in their diagnostic processes and facilitating the development of automated histopathological analysis systems.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110321"},"PeriodicalIF":7.0,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124316","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":"Machine learning in biofluid mechanics: A review of recent developments","authors":"Amirmohammad Sattari","doi":"10.1016/j.compbiomed.2025.110410","DOIUrl":"10.1016/j.compbiomed.2025.110410","url":null,"abstract":"<div><div>This review paper comprehensively examines recent advancements in machine learning (ML) applications within biofluid mechanics, with a targeted focus on enabling clinically actionable diagnostics and simulations. It demonstrates how ML, and in particular physics-informed ML methods, are used to enhance the analysis and understanding of intricate biofluid dynamics. The review systematically analyzes various ML techniques, detailing their strengths and limitations in modeling biofluid behaviors. By integrating physics-informed ML methods, such as Physics-Informed Neural Networks (PINNs), this work addresses critical challenges in translating complex biofluid dynamics into practical clinical tools. Differentiating itself from previous literature, this review not only summarizes current methods but also proposes potential solutions—including data augmentation, transfer learning, and hybrid modeling approaches (e.g., PINNs)—to overcome challenges related to limited datasets and the integration of complex physics. The review emphasizes ML's ability to enhance diagnostic accuracy, enable personalized treatment strategies, and accelerate computational simulations for applications like cardiovascular disease detection and respiratory disorder diagnosis, with findings showing that ML-driven approaches can reduce diagnostic errors by up to 30 % in cardiovascular applications and improve early detection rates in metabolomics-based diagnostics. Findings indicate that while ML techniques have significantly improved predictive capabilities in biofluid dynamics, challenges such as data scarcity and multi-scale physical integration remain critical. By outlining strategies to bridge the gap between ML advancements and clinical implementation, this review provides a robust framework for future research aimed at integrating ML with biofluid mechanics to revolutionize healthcare delivery. The paper concludes by identifying future research directions aimed at further integrating ML with domain-specific physical insights to achieve more reliable and accurate biofluid models.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"193 ","pages":"Article 110410"},"PeriodicalIF":7.0,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124320","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}