Computers in biology and medicine最新文献

{"title":"SegElegans: Instance segmentation using dual convolutional recurrent neural network decoder in Caenorhabditis elegans microscopic images","authors":"Pablo E. Layana Castro ,&nbsp;Konstantinos Kounakis ,&nbsp;Antonio García Garví ,&nbsp;Ilias Gkikas ,&nbsp;Ioannis Tsiamantas ,&nbsp;Nektarios Tavernarakis ,&nbsp;Antonio-José Sánchez-Salmerón","doi":"10.1016/j.compbiomed.2025.110012","DOIUrl":"10.1016/j.compbiomed.2025.110012","url":null,"abstract":"<div><div><em>Caenorhabditis elegans</em> is a great model for exploring organismal, cellular, and subcellular biology through optical and fluorescence microscopy, with its research applications steadily expanding. However, manual processing of numerous microscopic images is prone to errors and demands significant labor due to worms tendency to touch or cluster with each other. Here, we present a new system for segmenting whole-body instances of <em>Caenorhabditis elegans</em> in microscopic images (referred to as SegElegans), employing a combination of neural network architecture and conventional image processing techniques. Our method effectively overcomes previous challenges and resolves many instances of contact and overlap between worms in highly populated images in a timely manner. The results obtained show an average Intersection over Union value of 96.3% per worm and an average improvement of 6% over other existing methods for automated analysis of worm images. SegElegns is a user-friendly application for <em>Caenorhabditis elegans</em> segmentation that will benefit whole-worm phenotypic screenings essential for studying development, behavior, aging, and disease.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110012"},"PeriodicalIF":7.0,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687544","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HistoMSC: Density and topology analysis for AI-based visual annotation of histopathology whole slide images","authors":"Zahoor Ahmad ,&nbsp;Khaled Al-Thelaya ,&nbsp;Mahmood Alzubaidi ,&nbsp;Faaiz Joad ,&nbsp;Nauman Ullah Gilal ,&nbsp;William Mifsud ,&nbsp;Sabri Boughorbel ,&nbsp;Giovanni Pintore ,&nbsp;Enrico Gobbetti ,&nbsp;Jens Schneider ,&nbsp;Marco Agus","doi":"10.1016/j.compbiomed.2025.109991","DOIUrl":"10.1016/j.compbiomed.2025.109991","url":null,"abstract":"<div><div>We introduce an end-to-end framework for the automated visual annotation of histopathology whole slide images. Our method integrates deep learning models to achieve precise localization and classification of cell nuclei with spatial data aggregation to extend classes of sparsely distributed nuclei across the entire slide. We introduce a novel and cost-effective approach to localization, leveraging a U-Net architecture and a ResNet-50 backbone. The performance is boosted through color normalization techniques, helping achieve robustness under color variations resulting from diverse scanners and staining reagents. The framework is complemented by a YOLO detection architecture, augmented with generative methods. For classification, we use context patches around each nucleus, fed to various deep architectures. Sparse nuclei-level annotations are then aggregated using kernel density estimation, followed by color-coding and isocontouring. This reduces visual clutter and provides per-pixel probabilities with respect to pathology taxonomies. Finally, we use Morse–Smale theory to generate abstract annotations, highlighting extrema in the density functions and potential spatial interactions in the form of abstract graphs. Thus, our visualization allows for exploration at scales ranging from individual nuclei to the macro-scale. We tested the effectiveness of our framework in an assessment by six pathologists using various neoplastic cases. Our results demonstrate the robustness and usefulness of the proposed framework in aiding histopathologists in their analysis and interpretation of whole slide images.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 109991"},"PeriodicalIF":7.0,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687545","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Novel classification of brain vascular tortuosity measures: A systematic review","authors":"Sandra Bernaus ,&nbsp;Júlia Romagosa ,&nbsp;Christian Mata ,&nbsp;Christian Stephan-Otto ,&nbsp;Raúl Benítez ,&nbsp;Arnau Valls-Esteve ,&nbsp;Josep Munuera","doi":"10.1016/j.compbiomed.2025.109990","DOIUrl":"10.1016/j.compbiomed.2025.109990","url":null,"abstract":"<div><div>Given the absence of a standardized measure for evaluating tortuosity in cerebrovascular images, our investigation focuses on the methods used to estimate vascular tortuosity over the past decade. The main purpose is to create a useful, easily accessible guide to tortuosity estimation methods for brain researchers and clinicians. We conducted a systematic literature review in PUBMED and Scopus from 2013 to 2023 for tortuosity index (TI) analysis of human cerebrovascular images providing either quantitative or qualitative tortuosity measures. A total of 111 articles reporting TI measures were identified, in which 16 different TI were used to analyze 29 different diseases in Magnetic Resonance Angiography (MRA), Computed Tomography Angiography (CTA), Digital Subtraction Angiography (DSA), Ultrasound images (US), and other Magnetic Resonance Imaging (MRI) sequences. A novel categorization of tortuosity indices is suggested, based on the nature of the metrics. This classification comprises four categories: morphological-based, ratio distance-based, trigonometrical-based, and curvature-based methods. A TI Metric guide is proposed to facilitate the selection of the optimal TI for each use case. Our results show that Distance Metric (DM) is the most used, simple, and versatile method of capturing tortuous patterns, making it a preferred choice among researchers studying different disease contexts. Conversely, healthcare practitioners often prefer the Weibel and Fields tortuosity metric due to its categorical output, which offers a simplified and readily interpretable classification that complements clinical decision-making processes effectively.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 109990"},"PeriodicalIF":7.0,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143673511","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":"Computational profiling of molecular biomarkers in congenital disorders of glycosylation Type-I and binding analysis of Ginkgolide A with P4HB","authors":"Muhammad Rahiyab,&nbsp;Ishaq Khan,&nbsp;Syed Shujait Ali,&nbsp;Zahid Hussain,&nbsp;Shahid Ali,&nbsp;Arshad Iqbal","doi":"10.1016/j.compbiomed.2025.110042","DOIUrl":"10.1016/j.compbiomed.2025.110042","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;Aims&lt;/h3&gt;&lt;div&gt;Congenital disorders of glycosylation (CDG) comprise a diverse group of genetic diseases characterized by aberrant glycosylation that leads to severe multi-systematic effects. Despite advancements in understanding the underlying molecular mechanisms, curative options remain limited. This study employed computational methods to identify key molecular biomarkers for CDG-I and examine the pharmacological effects of Ginkgolide A (GA), a potent bioactive natural compound.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Methods&lt;/h3&gt;&lt;div&gt;We analyzed the GSE8440 microarray dataset to discover differentially expressed genes (DEGs) in patients compared to healthy individuals with CDG-I utilizing GEO2R. Functional enrichments, including gene ontologies (GO) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analyses, were conducted to contextualize the biological mechanisms and molecular signatures involved in CDG-I (Congenital Disorders of Glycosylation Type-1). The protein-protein interaction (PPI) network for DEGs was constructed using the STRING database, and the central hub genes within the PPI network were identified using Cytohubba. Furthermore, the 3D structure of the top hub gene (P4HB) was predicted by using the Robetta server. The CASTp was employed to evaluate the active sites. Molecular docking of P4HB with GA was carried out to investigate the binding affinity using the PyRx tool, and the stability of the docked complex was validated through MD simulation. The pharmacokinetics, toxicity, and bioactivity score of GA were comprehensively assessed using SwissADME, ProTox-II, and Molinspiration.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Results&lt;/h3&gt;&lt;div&gt;Our findings indicated 247 significant DEGs, including 146 up-regulated and 101 down-regulated genes. GO and KEGG pathway analyses confirmed that the up-regulated and hub genes were strongly associated with protein folding, glycoprotein processing in the endoplasmic reticulum, and endoplasmic reticulum stress (ER) pathways. P4HB emerged as the top hub gene in CDG-I, playing a significant role in protein folding and ER stress. The 3D structure of P4HB was refined and validated, achieving 95.8 % residues in the most favored region of the Ramachandran plot, with an overall quality of 92.97 %. The CASTp server predicted the largest active site with an area of 2243.660 Å&lt;sup&gt;2&lt;/sup&gt; and a volume of 3236.584 Å&lt;sup&gt;3&lt;/sup&gt;. Molecular docking revealed that GA has a strong binding affinity with P4HB (−8.9 kcal/mol). The ADME (Absorption, Distribution, Metabolism, Excretion) and toxicity assessments confirmed promising drug-like characteristics, excellent bioavailability, and minimal toxicity risk.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Conclusion&lt;/h3&gt;&lt;div&gt;This study emphasizes GA as a potential treatment possibility option to alleviated CDG-I pathology by targeting protein misfolding and ER stress, which are fundamental aspects of the disease. Additionally, our findings indicate that P4HB is a critical molecular target in CDG-I. These results p","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110042"},"PeriodicalIF":7.0,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143673507","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 Bayesian Belief Network model for the estimation of risk of cardiovascular events in subjects with type 1 diabetes","authors":"Ornella Moro ,&nbsp;Inger Torhild Gram ,&nbsp;Maja-Lisa Løchen ,&nbsp;Marit B. Veierød ,&nbsp;Ana Maria Wägner ,&nbsp;Giovanni Sebastiani","doi":"10.1016/j.compbiomed.2025.109967","DOIUrl":"10.1016/j.compbiomed.2025.109967","url":null,"abstract":"<div><h3>Objectives:</h3><div>Cardiovascular diseases (CVDs) represent a major risk for people with type 1 diabetes (T1D). Our aim here is to develop a new methodology that overcomes some of the problems and limitations of existing risk calculators. First, they are rarely tailored to people with T1D and, in general, they do not deal with missing values for any risk factor. Moreover, they do not take into account information on risk factors dependencies, which is often available from medical experts.</div></div><div><h3>Method:</h3><div>This study introduces a Bayesian Belief Network (BBN) model to quantify CVD risk in individuals with T1D. The developed methodology is applied to a large T1D dataset and its performances are assessed. A simulation study is also carried out to quantify the parameter estimation properties.</div></div><div><h3>Results:</h3><div>The performances of individual risk estimation, as measured by the area under the ROC curve and by the C-index, are about 0.75 for both real and simulated data with comparable sample sizes.</div></div><div><h3>Conclusions:</h3><div>We observe a good predictive ability of the proposed methodology with accurate parameter estimation. The BBN approach takes into account causal relationships between variables, providing a comprehensive description of the system. This makes it possible to derive useful tools for optimising intervention.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 109967"},"PeriodicalIF":7.0,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143673505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing skin disease classification leveraging transformer-based deep learning architectures and explainable AI","authors":"Jayanth Mohan ,&nbsp;Arrun Sivasubramanian ,&nbsp;Sowmya V. ,&nbsp;Vinayakumar Ravi","doi":"10.1016/j.compbiomed.2025.110007","DOIUrl":"10.1016/j.compbiomed.2025.110007","url":null,"abstract":"<div><div>Skin diseases affect over a third of the global population, yet their impact is often underestimated. Automating the classification of these diseases is essential for supporting timely and accurate diagnoses. This study leverages Vision Transformers, Swin Transformers, and DinoV2, introducing DinoV2 for the first time in dermatology tasks. On a 31-class skin disease dataset, DinoV2 achieves state-of-the-art results with a test accuracy of 96.48 ± 0.0138% and an F1-Score of 97.27%, marking a nearly 10% improvement over existing benchmarks. The robustness of DinoV2 is further validated on the HAM10000 and Dermnet datasets, where it consistently surpasses prior models. Comparative analysis also includes ConvNeXt and other CNN architectures, underscoring the benefits of transformer models. Additionally, explainable AI techniques like GradCAM and SHAP provide global heatmaps and pixel-level correlation plots, offering detailed insights into disease localization. These complementary approaches enhance model transparency and support clinical correlations, assisting dermatologists in accurate diagnosis and treatment planning. This combination of high performance and clinical relevance highlights the potential of transformers, particularly DinoV2, in dermatological applications.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110007"},"PeriodicalIF":7.0,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143673508","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":"Intersection-based slice motion estimation for fetal brain imaging","authors":"Chloe Mercier ,&nbsp;Sylvain Faisan ,&nbsp;Alexandre Pron ,&nbsp;Nadine Girard ,&nbsp;Guillaume Auzias ,&nbsp;Thierry Chonavel ,&nbsp;François Rousseau","doi":"10.1016/j.compbiomed.2025.110005","DOIUrl":"10.1016/j.compbiomed.2025.110005","url":null,"abstract":"<div><div>Fetal MRI offers a broad spectrum of applications, including the investigation of fetal brain development and facilitation of early diagnosis. However, image quality is often compromised by motion artifacts arising from both maternal and fetal movement. To mitigate these artifacts, fetal MRI typically employs ultrafast acquisition sequences. This results in the acquisition of three (or more) orthogonal stacks along different spatial axes. Nonetheless, inter-slice motion can still occur. If left uncorrected, such motion can introduce artifacts in the reconstructed 3D volume. Existing motion-correction approaches often rely on a two-step iterative process involving registration followed by reconstruction. They tend to detect and remove a large number of misaligned slices, resulting in poor reconstruction quality. This paper proposes a novel reconstruction-independent method for motion correction. Our approach benefits from the intersection of orthogonal slices and estimates motion for each slice by minimizing the difference between the intensity profiles along their intersections. To address potential misalignments, we present an innovative machine learning-based classifier for identifying misaligned slices. The parameters of these slices are then corrected using a multistart optimization approach. Quantitative evaluation on simulated datasets demonstrates very low registration errors. Qualitative analysis on real data further highlights the effectiveness of our approach compared to state-of-the-art methods.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110005"},"PeriodicalIF":7.0,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143654652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Identification of lncRNA associated with the SERPINE1 gene in colorectal cancer through TGF-β pathway","authors":"Ghazale Habibzadeh ,&nbsp;Khatere Mokhtari ,&nbsp;Masoomeh Heshmati ,&nbsp;Siamak Salimy ,&nbsp;Zhiqiang Mei ,&nbsp;Maliheh Entezari ,&nbsp;Mehrdad Hashemi ,&nbsp;Junjiang Fu ,&nbsp;Mazaher Maghsoudloo","doi":"10.1016/j.compbiomed.2025.110037","DOIUrl":"10.1016/j.compbiomed.2025.110037","url":null,"abstract":"<div><div>Colorectal cancer (CRC) is a common cancer type which develops due to intricate molecular processes, the Transforming Growth Factor-beta (TGF-β) pathway a role in progression. This study investigates the immunological functions of <em>SERPINE1</em> and its interaction with long non-coding RNA (lncRNA) <em>LINC01705</em> within the TGF-β pathway, aiming to identify novel therapeutic targets for CRC. We hypothesized that <em>LINC01705</em> modulates <em>SERPINE1</em> expression, thereby influencing CRC progression and immune response. To test this hypothesis, we employed bioinformatics analysis of the TCGA-COAD dataset and experimental validation through RT-qPCR. Our findings revealed a significant upregulation of <em>SERPINE1</em> in CRC, with nine interacting proteins involved in CRC-related processes identified through coexpression network analysis. Moreover, our findings revealed a high prevalence of mutations in <em>SERPINE1</em>, highlighting its potential as a target for immunotherapy. Additionally, we identified a strong correlation between <em>LINC01705</em> and <em>SERPINE1</em>, with experimental validation confirming their concurrent upregulation in CRC tissues. These results highlight the importance of the <em>SERPINE1/LINC01705</em> axis as a novel candidate that influence the TGF-β pathway, offering new insights into CRC pathogenesis and providing potential targets for diagnosis and immunotherapy.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110037"},"PeriodicalIF":7.0,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143654687","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":"The proteomic code: Novel amino acid residue pairing models “encode” protein folding and protein-protein interactions","authors":"Tareq Hameduh ,&nbsp;Andrew D. Miller ,&nbsp;Zbynek Heger ,&nbsp;Yazan Haddad","doi":"10.1016/j.compbiomed.2025.110033","DOIUrl":"10.1016/j.compbiomed.2025.110033","url":null,"abstract":"<div><div>Recent advances in protein 3D structure prediction using deep learning have focused on the importance of amino acid residue-residue connections (<em>i.e.</em>, pairwise atomic contacts) for accuracy at the expense of mechanistic interpretability. Therefore, we decided to perform a series of analyses based on an alternative framework of residue-residue connections making primary use of the TOP2018 dataset. This framework of residue-residue connections is derived from amino acid residue pairing models both historic and new, all based on genetic principles complemented by relevant biophysical principles. Of these pairing models, three new models (named the GU, Transmuted and Shift pairing models) exhibit the highest observed-over-expected ratios and highest correlations in statistical analyses with various intra- and inter-chain datasets, in comparison to the remaining models. In addition, these new pairing models are universally frequent across different connection ranges, secondary structure connections, and protein sizes. Accordingly, following further statistical and other analyses described herein, we have come to a major conclusion that all three pairing models together could represent the basis of a universal proteomic code (second genetic code) sufficient, in and of itself, to “encode” for both protein folding mechanisms and protein-protein interactions.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 110033"},"PeriodicalIF":7.0,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143654688","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Sequential classification approach for enhancing the assessment of cardiac autonomic neuropathy","authors":"Moustafa Abdelwanis ,&nbsp;Karim Moawad ,&nbsp;Shahmir Mohammed ,&nbsp;Ammar Hummieda ,&nbsp;Shayaan Syed ,&nbsp;Maher Maalouf ,&nbsp;Herbert F. Jelinek","doi":"10.1016/j.compbiomed.2025.109999","DOIUrl":"10.1016/j.compbiomed.2025.109999","url":null,"abstract":"<div><div>Cardiac autonomic neuropathy (CAN) is a progressive condition associated with chronic diseases like diabetes, requiring regular reviews. Current CAN diagnostic methods are often time-consuming and lack precision. This study presents a novel, two-stage classification model designed to improve CAN diagnostic efficiency. Using a dataset of 1335 patient entries, including inflammatory markers and autonomic function tests (CARTs), the model first classifies patients based on six inflammatory markers– Interleukin-6 (IL-6), C-reactive protein (CRP), Interleukin-1 beta (IL-1beta), Interleukin-10 (IL-10), Monocyte Chemoattractant Protein-1 (MCP-1), and Insulin-like growth factor-1 (IGF-1). In this initial stage, the model achieves 0.893 accuracy for 31.46% of cases in the three-class CAN model at a 0.80 threshold. For cases requiring further assessment, the second stage incorporates CARTs, improving overall accuracy to 0.933. Notably, 98.87% of cases are accurately classified using only a subset of CARTs, with just 1.12% needing all five tests. Additionally, we developed a web application that utilizes Shapley plots to visualize and explain the contribution of each marker, facilitating interpretation for clinical use. This two-stage approach underscores the diagnostic relevance of inflammatory markers, providing clinicians with a streamlined, resource-efficient tool for timely CAN diagnosis and intervention.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"190 ","pages":"Article 109999"},"PeriodicalIF":7.0,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143654651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}