Computers in biology and medicine最新文献

{"title":"Temporal validation of machine learning models for pre-eclampsia prediction using routinely collected maternal characteristics: A validation study","authors":"Sofonyas Abebaw Tiruneh ,&nbsp;Daniel Lorber Rolnik ,&nbsp;Helena Teede ,&nbsp;Joanne Enticott","doi":"10.1016/j.compbiomed.2025.110183","DOIUrl":"10.1016/j.compbiomed.2025.110183","url":null,"abstract":"<div><h3>Background</h3><div>Pre-eclampsia (PE) contributes to more than one-fourth of all maternal deaths and half a million newborn deaths worldwide every year. Early screening and interventions can reduce PE incidence and related complications. We aim to 1) temporally validate three existing models (two machine learning (ML) and one logistic regression) developed in the same region and 2) compare the performances of the validated ML models with the logistic regression model in PE prediction. This work addresses a gap in the literature by undertaking a comprehensive evaluation of existing risk prediction models, which is an important step to advancing this field.</div></div><div><h3>Methods</h3><div>We obtained a dataset of routinely collected antenatal data from three maternity hospitals in South-East Melbourne, Australia, extracted between July 2021 and December 2022. We temporally validated three existing models: extreme gradient boosting (XGBoost, ‘model 1’), random forest (‘model 2’) ML models, and a logistic regression model (‘model 3’). Area under the receiver-operating characteristic (ROC) curve (AUC) was evaluated discrimination performance, and calibration was assessed. The AUCs were compared using the ‘bootstrapping’ test.</div></div><div><h3>Results</h3><div>The temporal evaluation dataset consisted of 12,549 singleton pregnancies, of which 431 (3.43 %, 95 % confidence interval (CI) 3.13–3.77) developed PE. The characteristics of the temporal evaluation dataset were similar to the original development dataset. The XGBoost ‘model 1’ and the logistic regression ‘model 3’ exhibited similar discrimination performance with an AUC of 0.75 (95 % CI 0.73–0.78) and 0.76 (95 % CI 0.74–0.78), respectively. The random forest ‘model 2’ showed a discrimination performance of AUC 0.71 (95 % CI 0.69–0.74). Model 3 showed perfect calibration performance with a slope of 1.02 (95 % CI 0.92–1.12). Models 1 and 2 showed a calibration slope of 1.15 (95 % CI 1.03–1.28) and 0.62 (95 % CI 0.54–0.70), respectively. Compared to the original development models, the temporally validated models 1 and 3 showed stable discrimination performance, whereas model 2 showed significantly lower discrimination performance. Models 1 and 3 showed better clinical net benefits between 3 % and 22 % threshold probabilities than default strategies.</div></div><div><h3>Conclusions</h3><div>During temporal validation of PE prediction models, logistic regression and XGBoost models exhibited stable prediction performance; however, both ML models did not outperform the logistic regression model. To facilitate insights into interpretability and deployment, the logistic regression model could be integrated into routine practice as a first-step in a two-stage screening approach to identify a higher-risk woman for further second stage screening with a more accurate test.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110183"},"PeriodicalIF":7.0,"publicationDate":"2025-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823628","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":"BCOR-rearranged sarcomas: In silico insights into altered domains and BCOR interactions","authors":"Kristóf Madarász ,&nbsp;János András Mótyán ,&nbsp;Yi-Che Chang Chien ,&nbsp;Judit Bedekovics ,&nbsp;Szilvia Lilla Csoma ,&nbsp;Gábor Méhes ,&nbsp;Attila Mokánszki","doi":"10.1016/j.compbiomed.2025.110144","DOIUrl":"10.1016/j.compbiomed.2025.110144","url":null,"abstract":"<div><div>BCOR (BCL-6 corepressor) rearranged small round cell sarcoma (BRS) represents an uncommon soft tissue malignancy, frequently characterized by the <em>BCOR</em>::<em>CCNB3</em> fusion. Other noteworthy fusions include <em>BCOR</em>::<em>MAML3</em>, <em>BCOR</em>::<em>CLGN</em>, <em>BCOR</em>::<em>MAML1</em>, <em>ZC3H7B</em>::<em>BCOR</em>, <em>KMT2D</em>::<em>BCOR</em>, <em>CIITA</em>::<em>BCOR</em>, <em>RTL9</em>::<em>BCOR</em>, and <em>AHR</em>::<em>BCOR</em>. The <em>BCOR</em> gene plays a pivotal role in the Polycomb Repressive Complex 1 (PRC1), essential for histone modification and gene silencing. It interfaces with the Polycomb group RING finger homolog (PCGF1). This study employed comprehensive <em>in silico</em> methodologies to investigate the structural and functional effects of <em>BCOR</em> fusion events in BRS. The analysis revealed significant alterations in the domain architecture of BCOR, which resulted in the loss of <em>BCL6</em>-regulated transcriptional repression. Furthermore, IUPred3 prediction indicated a significant increase in disorder in the C-terminal regions of the BCOR in the fusion proteins. A detailed analysis of the physicochemical properties by ProtParam revealed a decrease in isoelectric point, stability, and hydrophobicity. The analysis of protein structures predicted by AlphaFold3 using the PRODIGY algorithm revealed statistically significant deviations in binding affinities for BCOR-PCGF1 dimers and a non-canonical PRC1 variant tetramer compared to the wild-type BCOR. The findings provide a comprehensive summary and elucidation of the fusion proteome associated with BRS, suggesting a substantial impact on the stability and functionality of the fusion proteins, thereby contributing to the oncogenic mechanisms underlying BRS. In this study, we provide the first compilation and comparative analysis of the known BCOR fusions of BRS and introduce a new <em>in silico</em> approach to enhance a better understanding of the molecular basis of BRS.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110144"},"PeriodicalIF":7.0,"publicationDate":"2025-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823627","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 the frontier of artificial intelligence on emerging technologies to redefine cancer diagnosis and care","authors":"Akanksha Vyas ,&nbsp;Krishan Kumar ,&nbsp;Ayushi Sharma ,&nbsp;Damini Verma ,&nbsp;Dhiraj Bhatia ,&nbsp;Nitin Wahi ,&nbsp;Amit K. Yadav","doi":"10.1016/j.compbiomed.2025.110178","DOIUrl":"10.1016/j.compbiomed.2025.110178","url":null,"abstract":"<div><h3>Background</h3><div>Artificial Intelligence (AI) is capable of revolutionizing cancer therapy and advancing precision oncology <em>via</em> integrating genomics data and digitized health information. AI applications show promise in cancer prediction, prognosis, and treatment planning, particularly in radiomics, deep learning, and machine learning for early cancer diagnosis. However, widespread adoption requires comprehensive data and clinical validation. While AI has demonstrated advantages in treating common malignancies like lung and breast cancers, challenges remain in managing rare tumors due to limited datasets. AI's role in processing multi-omics data and supporting precision oncology decision-making is critical as genetic and health data become increasingly digitized.</div></div><div><h3>Method</h3><div>This review article presents current knowledge on AI and associated technologies, which are being utilized in the diagnosis and therapy of cancer. The applications of AI in radiomics, deep learning, and machine learning for cancer screening and treatment planning are examined. The study also explores the capabilities and limitations of predictive AI in diagnosis and prognosis, as well as generative AI, such as advanced chatbots, in patient and provider interactions.</div></div><div><h3>Results</h3><div>AI can improve the early diagnosis and treatment of high-incidence cancers like breast and lung cancer. However, its application in rare cancers is limited by insufficient data for training and validation. AI can effectively process large-scale multi-omics data from DNA and RNA sequencing, enhancing precision oncology. Predictive AI aids in risk assessment and prognosis, while generative AI tools improve patient-provider communication. Despite these advancements, further research and technological progress are needed to overcome existing challenges.</div></div><div><h3>Conclusions</h3><div>AI holds transformative potential for cancer therapy, particularly in precision oncology, early detection, and personalized treatment planning. However, challenges such as data limitations in rare cancers, the need for clinical validation, and regulatory considerations must be addressed. Future advancements in AI could significantly improve decision-support systems in oncology, ultimately enhancing patient care and quality of life. The review highlights both the opportunities and obstacles in integrating AI into cancer diagnostics and therapeutics, calling for continued research and regulatory oversight.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110178"},"PeriodicalIF":7.0,"publicationDate":"2025-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823512","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":"An integrative study on the effects of Lingguizhugan decoction in treating Alzheimer's disease rats through modulation of multiple pathways involving various components","authors":"Zhe Wang ,&nbsp;Hao-tian Zhang ,&nbsp;Si-yue Li ,&nbsp;Xiu-ping Song ,&nbsp;Chong-zhen Shi ,&nbsp;Ye-wen Zhang ,&nbsp;Fei Han","doi":"10.1016/j.compbiomed.2025.110149","DOIUrl":"10.1016/j.compbiomed.2025.110149","url":null,"abstract":"<div><h3>Objective</h3><div>To explore the active components and mechanisms of Lingguizhugan decoction (LGZGD) in the treatment of Alzheimer's disease (AD) through an integrated approach.</div></div><div><h3>Methods</h3><div>The active components of LGZGD in rat serum were identified using HPLC-FTICR MS. Network pharmacology and molecular docking analyses were conducted, and their findings were validated using an Aβ_1-42-induced AD rat model.</div></div><div><h3>Results</h3><div>Twenty-four active components and 324 common targets were identified and used to construct the networks. KEGG pathway enrichment analysis linked key target genes with MAPK, Rap1, and NF-κB signaling pathways. Molecular docking results indicated that three key targets (IL-6, TNF, and EGFR) and 10 core components are closely associated with LGZGD in the treatment of AD. LGZGD improved the spatial learning and memory abilities of AD rats. LGZGD reduced neuronal damage and increased the number of neurons in the cortex and hippocampal CA1 region of AD rats. LGZGD decreased Aβ_1-42 expression in the rat hippocampus, alleviated oxidative stress in AD rats, and decreased TNF-α, IL-6, IL-1β, and HMGB1 levels in the cerebral cortical tissue. LGZGD markedly decreased Iba-1 and iNOS expression and increased CD206 levels to inhibit M1 activation and promote M2 activation. LGZGD increased the expression of p-GSK-3β, ERK, and p-ERK, while decreasing the expression of p-Tau, IKKβ, p-IκBα, p-p65, p-p38, and p-JNK in the hippocampus of AD rats.</div></div><div><h3>Conclusion</h3><div>LGZGD treats AD by modulating targets like IL-6, TNF, MAPK3, and BCL2, thereby alleviating cognitive impairments in rats. Its neuroprotective effects in treating AD are mediated through the NF-κB/MAPK signaling pathways.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110149"},"PeriodicalIF":7.0,"publicationDate":"2025-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823625","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 novel smart guidewire with an integrated hemodynamic sensor for central catheter placement: Design and simulation","authors":"Huiying Huang ,&nbsp;Peiming Lu ,&nbsp;Minglu Zhong ,&nbsp;Handong Ouyang ,&nbsp;Shengzhao Lin","doi":"10.1016/j.compbiomed.2025.110139","DOIUrl":"10.1016/j.compbiomed.2025.110139","url":null,"abstract":"<div><h3>Objective</h3><div>We analyzed the differences in hemodynamic patterns along the central venous catheterization pathway and constructed a sensor-at-tip guidewire for real-time detection of temperature field changes related to hemodynamic patterns. The design was verified using COSMOL simulation and <em>in vitro</em> simulation tests to evaluate its potential application as a tool to facilitate navigation during catheterization.</div></div><div><h3>Methods</h3><div>Differences in the hemodynamic modes in the central venous catheterization pathway led to changes in the temperature field created with a thermal source. A sensor-at-tip guidewire model was used to detect real-time changes in the temperature field during catheterization. By multiphysical coupling of temperature, heating power, thermistor, and fluid velocity fields, a simulation study based on the intrinsic characteristics of thermistor material winding springs was conducted, wherein the coupling relationship between the blood flow velocity (flow rate) and temperature transfer was obtained and the design was verified by simulation.</div></div><div><h3>Results</h3><div>Based on a multiphysics finite element simulation, the application of a thermal flow sensor composed of a thermistor and power resistor in central vein catheterization was verified. Theoretical calculations suggested that the thermal flow sensor can be composed of a conventional wire-wound spring or a commercially available, inexpensive, small-sized (01005 package) negative thermal coefficient resistor. This study provides a low-cost, portable, and real-time navigation solution for hemodynamic monitoring that is expected to have clinical applications.</div></div><div><h3>Conclusion</h3><div>The sensitivity and resolution of this design met the requirements of difference analysis for heating power vs. temperature fields as well as hemodynamic changes vs. temperature fields, indicating potential applications in navigation for central venous catheterization.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110139"},"PeriodicalIF":7.0,"publicationDate":"2025-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823626","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":"Uncovering hepatic transcriptomic and circulating proteomic signatures in MASH: A meta-analysis and machine learning-based biomarker discovery","authors":"Elena Cristina Rusu ,&nbsp;Helena Clavero-Mestres ,&nbsp;Mario Sánchez-Álvarez ,&nbsp;Marina Veciana-Molins ,&nbsp;Laia Bertran ,&nbsp;Pablo Monfort-Lanzas ,&nbsp;Carmen Aguilar ,&nbsp;Javier Camaron ,&nbsp;Teresa Auguet","doi":"10.1016/j.compbiomed.2025.110170","DOIUrl":"10.1016/j.compbiomed.2025.110170","url":null,"abstract":"<div><h3>Background</h3><div>Metabolic-associated steatohepatitis (MASH), the progressive form of metabolic-associated steatotic liver disease (MASLD), poses significant risks for liver fibrosis and cardiovascular complications. Despite extensive research, reliable biomarkers for MASH diagnosis and progression remain elusive. This study aimed to identify hepatic transcriptomic and circulating proteomic signatures specific to MASH, and to develop a machine learning-based biomarker discovery model.</div></div><div><h3>Methods</h3><div>A systematic review of RNA-Seq and proteomic datasets was conducted, retrieving 7 hepatic transcriptomics and 3 circulating proteomics studies, encompassing 483 liver samples and 169 serum/plasma samples, respectively. Differential gene and protein expression analyses were performed, and pathways were enriched using gene set enrichment analysis. A machine learning (ML) model was developed to identify MASH-specific biomarkers, utilizing biologically significant protein ratios.</div></div><div><h3>Key findings</h3><div>Hepatic transcriptomic analysis identified 5017 differentially expressed genes (DEGs), with significant enrichment of extracellular matrix (ECM) pathways. Serum proteomics revealed six differentially expressed proteins (DEPs), including complement-related proteins. Integration of transcriptomic and proteomic data highlighted the complement cascade as a key pathway in MASH, with discordant regulation between the liver and circulation. The ML-based biomarker discovery model, utilizing protein ratios, achieved an F1 scores of 0.83 and 0.64 in the training sets and 0.67 in an external validation set.</div></div><div><h3>Conclusion</h3><div>Our findings indicate ECM deregulation and complement system involvement in MASH progression. The novel ML model incorporating protein ratios offers a potential tool for MASH diagnosis. However, further refinement and validation across larger and more diverse cohorts is needed to generalize these results.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110170"},"PeriodicalIF":7.0,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820284","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":"Evolutionary insights into elongation factor G using AlphaFold and ancestral analysis","authors":"Shawonur Rahaman ,&nbsp;Jacob H. Steele ,&nbsp;Yi Zeng ,&nbsp;Shoujun Xu ,&nbsp;Yuhong Wang","doi":"10.1016/j.compbiomed.2025.110188","DOIUrl":"10.1016/j.compbiomed.2025.110188","url":null,"abstract":"<div><div>Elongation factor G (EF-G) is crucial for ribosomal translocation, a fundamental step in protein synthesis. Despite its indispensable role, the conformational dynamics and evolution of EF-G remain elusive. By integrating AlphaFold structural predictions with multiple sequence alignment (MSA)-based sequence analysis, we explored the conformational landscape, sequence-specific patterns, and evolutionary divergence of EF-G. We identified five high-confidence structural states of wild type (WT) EF-G, revealing broader conformational diversity than previously captured by experimental data. Phylogenetic analysis and MSA-embedded sequence patterns demonstrated that single-point mutations in the switch I loop modulate equilibrium between the two dominant conformational states, con1 and con2, which exhibit distinct functional specializations. Reconstructions of two ancestral EF-Gs revealed minimal GTPase activity and reduced translocase function in both forms, suggesting that robust translocase activity emerged after the divergence of con1 and con2. However, ancestral EF-Gs retained the fidelity of three-nucleotide translocation, underscoring the early evolutionary conservation of accurate mRNA movement. These findings establish a framework for understanding how conformational flexibility shapes EF-G function and specialization. Moreover, our computational pipeline can be extended to other translational GTPases, providing broader insights into the evolution of the translational machinery. This study highlights the power of AlphaFold-assisted structural analysis in revealing the mechanistic and evolutionary relationships involved in protein translation.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110188"},"PeriodicalIF":7.0,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820285","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":"Optimized multiple instance learning for brain tumor classification using weakly supervised contrastive learning","authors":"Kaoyan Lu ,&nbsp;Shiyu Lin ,&nbsp;Kaiwen Xue ,&nbsp;Duoxi Huang ,&nbsp;Yanghong Ji","doi":"10.1016/j.compbiomed.2025.110075","DOIUrl":"10.1016/j.compbiomed.2025.110075","url":null,"abstract":"<div><div>Brain tumors have a great impact on patients’ quality of life and accurate histopathological classification of brain tumors is crucial for patients’ prognosis. Multi-instance learning (MIL) has become the mainstream method for analyzing whole-slide images (WSIs). However, current MIL-based methods face several issues, including significant redundancy in the input and feature space, insufficient modeling of spatial relations between patches and inadequate representation capability of the feature extractor. To solve these limitations, we propose a new multi-instance learning with weakly supervised contrastive learning for brain tumor classification. Our framework consists of two parts: a cross-detection MIL aggregator (CDMIL) for brain tumor classification and a contrastive learning model based on pseudo-labels (PSCL) for optimizing feature encoder. The CDMIL consists of three modules: an internal patch anchoring module (IPAM), a local structural learning module (LSLM) and a cross-detection module (CDM). Specifically, IPAM utilizes probability distribution to generate representations of anchor samples, while LSLM extracts representations of local structural information between anchor samples. These two representations are effectively fused in CDM. Additionally, we propose a bag-level contrastive loss to interact with different subtypes in the feature space. PSCL uses the samples and pseudo-labels anchored by IPAM to optimize the performance of the feature encoder, which extracts a better feature representation to train CDMIL. We performed benchmark tests on a self-collected dataset and a publicly available dataset. The experiments show that our method has better performance than several existing state-of-the-art methods.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110075"},"PeriodicalIF":7.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143815318","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 machine learning approach to differentiate stage IV from stage I colorectal cancer","authors":"Naim Abu-Freha ,&nbsp;Zaid Afawi ,&nbsp;Miar Yousef ,&nbsp;Walid Alamor ,&nbsp;Noor Sanalla ,&nbsp;Simon Esbit ,&nbsp;Malik Yousef","doi":"10.1016/j.compbiomed.2025.110179","DOIUrl":"10.1016/j.compbiomed.2025.110179","url":null,"abstract":"<div><h3>Background and aim</h3><div>The stage at which Colorectal cancer (CRC) diagnosed is a crucial prognostic factor. Our study proposed a novel approach to aid in the diagnosis of stage IV CRC by utilizing supervised machine learning, analyzing clinical history, and laboratory values, comparing them with those of stage I CRC.</div></div><div><h3>Methods</h3><div>We conducted a respective study using patients diagnosed with stage I (n = 433) and stage IV CRC (n = 457). We employed supervised machine learning using random forest. The decision tree is used to visualize the model to identify key clinical and laboratory factors that differentiate between stage IV and stage I CRC.</div></div><div><h3>Results</h3><div>The decision tree classifier revealed that symptoms combined with laboratory values were critical predictors of stage IV CRC. Change in bowel habits was predictive for stage IV CRC among 14 of 22 patients (63 %). Weight loss, constipation, and abdominal pain in combination with different levels of carcinoembryonic antigen (CEA) were predictors for stage IV CRC. A CEA level higher than 260 was indicative for stage IV CRC in all observed patients (61 out of 61 patients). Additionally, a lower CEA level, in combination with hemoglobin, white blood cell count, and platelet count, also predicted stage IV CRC.</div></div><div><h3>Conclusions</h3><div>By applying a machine learning based approach, we identified symptoms and laboratory values (CEA, hemoglobin, white blood cell count, and platelet count), as crucial predictors for stage IV CRC diagnosis. This method holds potential for facilitating the diagnosis of stage IV CRC in clinical practice, even before imaging tests are conducted.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":"Article 110179"},"PeriodicalIF":7.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820277","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 computed tomography-based deep learning radiomics model for predicting the gender-age-physiology stage of patients with connective tissue disease-associated interstitial lung disease","authors":"Bingqing Long ,&nbsp;Rui Li ,&nbsp;Ronghua Wang ,&nbsp;Anyu Yin ,&nbsp;Ziyi Zhuang ,&nbsp;Yang Jing ,&nbsp;Linning E","doi":"10.1016/j.compbiomed.2025.110128","DOIUrl":"10.1016/j.compbiomed.2025.110128","url":null,"abstract":"<div><h3>Objectives</h3><div>To explore the feasibility of using a diagnostic model constructed with deep learning-radiomics (DLR) features extracted from chest computed tomography (CT) images to predict the gender-age-physiology (GAP) stage of patients with connective tissue disease-associated interstitial lung disease (CTD-ILD).</div></div><div><h3>Materials and methods</h3><div>The data of 264 CTD-ILD patients were retrospectively collected. GAP Stage I, II, III patients are 195, 56, 13 cases respectively. The latter two stages were combined into one group. The patients were randomized into a training set and a validation set. Single-input models were separately constructed using the selected radiomics and DL features, while DLR model was constructed from both sets of features. For all models, the support vector machine (SVM) and logistic regression (LR) algorithms were used for construction. The nomogram models were generated by integrating age, gender, and DLR features.</div></div><div><h3>Results</h3><div>The DLR model outperformed the radiomics and DL models in both the training set and the validation set. The predictive performance of the DLR model based on the LR algorithm was the best among all the feature-based models (AUC = 0.923). The comprehensive models had even greater performance in predicting the GAP stage of CTD-ILD patients. The comprehensive model using the SVM algorithm had the best performance of the two models (AUC = 0.951).</div></div><div><h3>Conclusion</h3><div>The DLR model extracted from CT images can assist in the clinical prediction of the GAP stage of CTD-ILD patients. A nomogram showed even greater performance in predicting the GAP stage of CTD-ILD patients.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"191 ","pages":""},"PeriodicalIF":7.0,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143806934","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}