Computers in biology and medicine最新文献

{"title":"Magnetization transfer MRI (MT-MRI) detects white matter damage beyond the primary site of compression in degenerative cervical myelopathy using a novel semi-automated analysis","authors":"Fauziyya Muhammad ,&nbsp;Kenneth A. Weber II ,&nbsp;Grace Haynes ,&nbsp;Lance Villeneuve ,&nbsp;Lonnie Smith ,&nbsp;Alaa Baha ,&nbsp;Sanaa Hameed ,&nbsp;Ali F. Khan ,&nbsp;Yasin Dhaher ,&nbsp;Todd Parrish ,&nbsp;Michael Rohan ,&nbsp;Zachary A. Smith","doi":"10.1016/j.compbiomed.2025.111083","DOIUrl":"10.1016/j.compbiomed.2025.111083","url":null,"abstract":"<div><div>Degenerative cervical myelopathy (DCM) is the leading cause of spinal cord disorder in adults, yet conventional MRI cannot detect microstructural damage beyond the compression site. Current application of magnetization transfer ratio (MTR), while promising, suffer from limited standardization, operator dependence, and unclear added value over traditional metrics such as cross-sectional area (CSA).</div><div>To address these limitations, we utilized our semi-automated analysis pipeline built on the Spinal Cord Toolbox (SCT) platform to automate MTR extraction. Our method integrates deep learning-based convolutional neural networks (CNNs) for spinal cord segmentation, vertebral labeling via the global curve optimization algorithm and PAM50 template registration to enable automated MTR extraction. Using the Generic Spine Protocol, we acquired 3T T2w- and MT-MRI images from 30 patients with DCM and 15 age-matched healthy controls (HC). We computed MTR and CSA at the maximal compression level (C5-C6) and a distant, uncompressed region (C2-C3). We extracted regional and tract-specific MTR using probabilistic maps in template space. Diagnostic accuracy was assessed with ROC analysis, and k-means clustering reveal patients subgroups based on neurological impairments. Correlation analysis assessed associations between MTR measures and DCM deficits.</div><div>Patients with DCM showed significant MTR reductions in both compressed and uncompressed regions (p &lt; 0.05). At C2-C3, MTR outperformed CSA (AUC 0.74 vs 0.69) in detecting spinal cord pathology. Tract-specific MTR were correlated with dexterity, grip strength, and balance deficits. Our reproducible, computationally robust pipeline links microstructural injury to clinical outcomes in DCM and provides a scalable framework for multi-site quantitative MRI analysis of the spinal cord.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111083"},"PeriodicalIF":6.3,"publicationDate":"2025-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145057106","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":"Multi-encoder self-adaptive hard attention network with maximum intensity projections for lung nodule segmentation","authors":"Muhammad Usman ,&nbsp;Azka Rehman ,&nbsp;Abd Ur Rehman ,&nbsp;Abdullah Shahid ,&nbsp;Tariq Mahmood Khan ,&nbsp;Imran Razzak ,&nbsp;Minyoung Chung ,&nbsp;Yeong-Gil Shin","doi":"10.1016/j.compbiomed.2025.111059","DOIUrl":"10.1016/j.compbiomed.2025.111059","url":null,"abstract":"<div><div>Accurate lung nodule segmentation is crucial for early-stage lung cancer diagnosis, as it can substantially enhance patient survival rates. Computed tomography (CT) images are widely employed for early diagnosis in lung nodule analysis. However, the heterogeneity of lung nodules, size diversity, and the complexity of the surrounding environment pose challenges for developing robust nodule segmentation methods. In this study, we propose an efficient end-to-end framework, the Multi-Encoder Self-Adaptive Hard Attention Network (MESAHA-Net), which consists of three encoding paths, an attention block, and a decoder block that assimilates CT slice patches with both forward and backward maximum intensity projection (MIP) images. This synergy affords a profound contextual understanding of lung nodules and also results in a deluge of features. To manage the profusion of features generated, we incorporate a self-adaptive hard attention mechanism guided by region of interest (ROI) masks centered on nodular regions, which MESAHA-Net autonomously produces. The network sequentially undertakes slice-by-slice segmentation, emphasizing nodule regions to produce precise three-dimensional (3D) segmentation. The proposed framework has been comprehensively evaluated on the Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) dataset, the largest publicly available dataset for lung nodule segmentation. The results demonstrate that our approach is highly robust across various lung nodule types, outperforming previous state-of-the-art techniques in terms of segmentation performance and computational complexity, making it suitable for real-time clinical implementation of artificial intelligence (AI)-driven diagnostic tools.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111059"},"PeriodicalIF":6.3,"publicationDate":"2025-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145057107","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":"Understanding clinical decision-making in traditional East Asian medicine through dimensionality reduction: An empirical investigation","authors":"Hyojin Bae ,&nbsp;Bongsu Kang ,&nbsp;Chang-Eop Kim","doi":"10.1016/j.compbiomed.2025.111081","DOIUrl":"10.1016/j.compbiomed.2025.111081","url":null,"abstract":"<div><div>This study examines the clinical decision-making processes in Traditional East Asian Medicine (TEAM) by reinterpreting pattern identification (PI) through the lens of dimensionality reduction. Focusing on the Eight Principle Pattern Identification (EPPI) system and utilizing empirical data from the <em>Shang-Han-Lun</em>, we explore the necessity and significance of prioritizing the Exterior-Interior pattern in diagnosis and treatment selection. We test three hypotheses: whether the Ext–Int dimension contains the most information about patient symptoms, represents the most abstract and generalizable symptom information, and facilitates the selection of appropriate herbal prescriptions. Employing quantitative measures (e.g., abstraction index, cross-conditional generalization performance, decision tree regression), our results demonstrate that the Exterior-Interior dimension represents the most abstract and generalizable symptom and herbal prescription information, contributing to the efficient mapping between diagnosis and treatment. This research provides an objective framework for understanding the cognitive processes underlying TEAM, bridging traditional medical practices with modern computational approaches. The findings offer insights into the development of AI-driven diagnostic tools in TEAM and conventional medicine, with the potential to advance clinical practice, education, and research.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111081"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044940","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":"Regional attention-enhanced vision transformer for accurate Alzheimer's disease classification using sMRI data","authors":"Alireza jomeiri ,&nbsp;Ahmad Habibizad Navin ,&nbsp;Mahboubeh Shamsi","doi":"10.1016/j.compbiomed.2025.111065","DOIUrl":"10.1016/j.compbiomed.2025.111065","url":null,"abstract":"<div><div>Alzheimer's disease (AD) poses a significant global health challenge, necessitating early and accurate diagnosis to enable timely intervention. Structural MRI (sMRI) is a key imaging modality for detecting AD-related brain atrophy, yet traditional deep learning models like convolutional neural networks (CNNs) struggle to capture complex spatial dependencies critical for AD diagnosis. This study introduces the Regional Attention-Enhanced Vision Transformer (RAE-ViT), a novel framework designed for AD classification using sMRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. RAE-ViT leverages regional attention mechanisms to prioritize disease-critical brain regions, such as the hippocampus and ventricles, while integrating hierarchical self-attention and multi-scale feature extraction to model both localized and global structural patterns. Evaluated on 1152 sMRI scans (255 AD, 521 MCI, 376 NC), RAE-ViT achieved state-of-the-art performance with 94.2 % accuracy, 91.8 % sensitivity, 95.7 % specificity, and an AUC of 0.96, surpassing standard ViTs (89.5 %) and CNN-based models (e.g., ResNet-50: 87.8 %). The model's interpretable attention maps align closely with clinical biomarkers (Dice: 0.89 hippocampus, 0.85 ventricles), enhancing diagnostic reliability. Robustness to scanner variability (92.5 % accuracy on 1.5T scans) and noise (92.5 % accuracy under 10 % Gaussian noise) further supports its clinical applicability. A preliminary multimodal extension integrating sMRI and PET data improved accuracy to 95.8 %. Future work will focus on optimizing RAE-ViT for edge devices, incorporating multimodal data (e.g., PET, fMRI, genetic), and exploring self-supervised and federated learning to enhance generalizability and privacy. RAE-ViT represents a significant advancement in AI-driven AD diagnosis, offering potential for early detection and improved patient outcomes.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111065"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044937","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":"Identification of drug targets in pan-drug resistant Acinetobacter baumannii via whole genome sequencing and subtractive genomics","authors":"Ayesha Tariq ,&nbsp;Ayaz Ahmed ,&nbsp;Ambrina Khatoon ,&nbsp;Asad Karim","doi":"10.1016/j.compbiomed.2025.111058","DOIUrl":"10.1016/j.compbiomed.2025.111058","url":null,"abstract":"<div><div>In this study, we report, to the best of our knowledge, the first complete genome sequence of a pan drug resistant (PDR<em>) Acinetobacter baumannii</em> strain (JRCGR-AK-AB-01) from Karachi, Pakistan. Strain JRCGR-AK-AB-01 exhibited a pan drug resistant phenotype, showing susceptibility only to polymyxin B and intermediate susceptibility to colistin. Hybrid genome sequencing using MinION long-reads and DNBSEQ short-reads revealed that the genome size of strain JRCGR-AK-AB-01 is 4.03 Mb. We identified that JRCGR-AK-AB-01 is closely related to other <em>A. baumannii</em> strains based on Average Amino Acid Identity (AAI), Genome-to-Genome Distance Calculator (GGDC), and Average Nucleotide Identity (ANI) analyses. Furthermore, pan-genome analysis revealed an open pan-genome, indicating frequent gene exchange. Subsequently, a subtractive genomics approach was employed to identify potential drug targets within the core genes that are essential, druggable, and non-homologous to both human proteins and gut microbiota. Finally, the selected genes were screened against the JRCGR-AK-AB-01 proteome to eliminate redundancies. Among these, NADP-dependent isocitrate dehydrogenase (IDH) was used for downstream analysis. Its structure was predicted via homology modeling and validated using different bioinformatics tools. Molecular docking and molecular dynamics (MD) simulations revealed that neomycin and paromomycin were the potent drugs against <em>Acinetobacter</em> spp. <em>In vitro</em> studies confirmed that neomycin (2.25 mg/mL) exhibited antimicrobial activity against the PDR strain of <em>A. baumannii</em>. Overall, this study defines genomic features and identifies potential therapeutic targets in PDR <em>A. baumannii</em>, thereby providing a foundation for future experimental validation and novel treatment strategies.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111058"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044941","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":"Skeletal muscle death from the perspective of electrical impedance as evidenced by experiment and numerical modelling","authors":"Rok Šmerc,&nbsp;Damijan Miklavčič,&nbsp;Samo Mahnič-Kalamiza","doi":"10.1016/j.compbiomed.2025.111073","DOIUrl":"10.1016/j.compbiomed.2025.111073","url":null,"abstract":"<div><div>Understanding the biophysical changes in skeletal muscle tissue during the minutes to hours post-excision or following irreversible damage is critical for biomedical applications and food processing. Muscle tissue, composed of myofibrillar and sarcoplasmic proteins, water, lipids, and connective tissue, forms a complex network of interactions that persists as it degrades post-mortem. This study investigates skeletal muscle death through ex vivo experimental measurements on porcine muscle, supported by a novel numerical model that builds the muscle up from individual fibres. Skeletal muscle tissue was found to exhibit strong anisotropy in electrical conductivity due to its structure. It demonstrates much lower conductivity perpendicular to muscle fibres compared to parallel with them owing to its limited plasma membrane conductivity. We explore how post-mortem changes, including increased membrane permeability during membrane decomposition, and external interventions like electroporation, alter these anisotropic properties. Our findings have implications for biomedicine, specifically treatments targeting muscle tissue, such as pulsed field ablation for cardiac arrhythmias, and characterisation of in-vitro engineered muscle tissues. In food production, the study informs applications of pulsed electric fields to modify meat structure and texture. By integrating experimental and theoretical approaches, this work provides new insights into the electrochemical and structural dynamics of skeletal muscle during and after death.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111073"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044939","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":"Nanoparticles and bioactive materials against COVID-19 and its variants: Hints from a computational-materials design perspective","authors":"Alexandre A. de Castro ,&nbsp;Letícia C. Assis ,&nbsp;Renan J.C. Appel ,&nbsp;Elaine F.F. da Cunha ,&nbsp;Eugenie Nepovimova ,&nbsp;Kamil Kuca ,&nbsp;Teodorico C. Ramalho ,&nbsp;Felipe A. La Porta","doi":"10.1016/j.compbiomed.2025.111046","DOIUrl":"10.1016/j.compbiomed.2025.111046","url":null,"abstract":"<div><div>Society currently faces many challenges caused by the coronavirus outbreak, known as SARS-CoV-2, and in addition, its variant strains tend to be still more aggressive. Therefore, there is an enormous need to accelerate the development of novel remediation techniques against SARS-CoV-2. A literature review focusing on key terms such as ‘COVID-19′, ‘SARS-CoV-2′, ‘pharmacotherapy’, ‘pandemic’, ‘nanotechnology’ and ‘computational-materials design’, accentuates the increased role played by <em>in silico</em> models in developing prevention, diagnosis, and the treatment strategies. In an attempt to help in the front line, computational repositioning of drugs has intensively been explored as a well-established strategy in preclinical research in order to discover an effective therapy for SARS-CoV-2 infection. Furthermore, computational-materials design—which integrates principles of materials science, physics, chemistry, and computer science—has emerged as an indispensable approach in the fight against COVID-19, accelerating the development of novel nanomaterials and bioactive compounds, and optimizing existing drugs for enhanced efficacy against the virus and its variants. Overall, in this review, we have demonstrated the vital role of computational-materials design strategies in diverse applications (such as diagnostics, vaccines, and treatments, as well as in understanding the fundamental mechanisms of the virus and its interactions with various advanced materials) to handle the current pandemic and pave the way toward future preparedness against emerging infectious disease outbreaks.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111046"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044946","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":"Impact of feature selection and feature engineering in prediction of cardiovascular diseases","authors":"Divya Yadav ,&nbsp;Deepika Rani ,&nbsp;Om Prakash Verma","doi":"10.1016/j.compbiomed.2025.111027","DOIUrl":"10.1016/j.compbiomed.2025.111027","url":null,"abstract":"<div><div>Heart disease remains a leading cause of death worldwide, presenting major challenges to public health. As a complex cardiovascular disorder (CVD), it often leads to life threatening complications such as heart attacks, strokes, and heart failure. Early and accurate diagnosis is essential for reducing fatality rates and ensuring better clinical results. While ML models have shown great potential in identifying cardiovascular diseases but their effectiveness heavily depends on optimal feature selection and engineering. This study presents a novel approach integrating feature selection and feature engineering techniques to enhance heart disease prediction using ML classifiers. Firstly, four key attributes were selected from a combined heart disease prediction dataset using RF model. Subsequently, a feature engineering technique have been applied to generate thirty six new features through basic arithmetic operations, strengthening dataset to improve predictive efficiency. These newly generated features have been utilized to train ML classifiers. To further enhance the classification accuracy, an ensemble learning approach based on soft voting have been applied to mitigate the impact of weaker classifiers. The effectiveness of the proposed methodology have been evaluated by comparing model performance with and without feature selection and engineering. The RF model achieved superior classification results, with accuracy, precision, recall, F1-score, AUC-ROC, and Jaccard score reaching 96.56%, 97.83%, 95.26%, 96.53%, 99.55%, 93.29% for RF model, respectively. Similarly, with feature engineering technique, the DT model attained 95.23%, 94.32%, 96.31%, 95.31%, 96.14%, 91.04%, respectively. Notably, RF and DT models demonstrated superior performance incorporating feature selection and engineering techniques. The present study shows high disease prediction accuracy as compared to various feature selection techniques and also, shows the significance of feature engineering in enhancing ML models, enabling medical professionals to diagnose diseases more effectively and at earlier stages.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111027"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044942","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":"Sex classification from hand X-ray images in pediatric patients: How zero-shot Segment Anything Model (SAM) can improve medical image analysis","authors":"Ramón A. Mollineda ,&nbsp;Karel Becerra ,&nbsp;Boris Mederos","doi":"10.1016/j.compbiomed.2025.111060","DOIUrl":"10.1016/j.compbiomed.2025.111060","url":null,"abstract":"<div><div>The potential to classify sex from hand data is a valuable tool in both forensic and anthropological sciences. This work presents possibly the most comprehensive study to date of sex classification from hand X-ray images. The research methodology involves a systematic evaluation of zero-shot Segment Anything Model (SAM) in X-ray image segmentation, a novel hand mask detection algorithm based on geometric criteria leveraging human knowledge (avoiding costly retraining and prompt engineering), the comparison of multiple X-ray image representations including hand bone structure and hand silhouette, a rigorous application of deep learning models and ensemble strategies, visual explainability of decisions by aggregating attribution maps from multiple models, and the transfer of models trained from hand silhouettes to sex prediction of prehistoric handprints. Training and evaluation of deep learning models were performed using the RSNA Pediatric Bone Age dataset, a collection of hand X-ray images from pediatric patients. Results showed very high effectiveness of zero-shot SAM in segmenting X-ray images, the contribution of segmenting before classifying X-ray images, hand sex classification accuracy above 95% on test data, and predictions from ancient handprints highly consistent with previous hypotheses based on sexually dimorphic features. Attention maps highlighted the carpometacarpal joints in the female class and the radiocarpal joint in the male class as sex discriminant traits. These findings are anatomically very close to previous evidence reported under different databases, classification models and visualization techniques.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111060"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044943","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 hybrid quorum sensing model for neurodynamic feature optimization in EEG-based Parkinson’s disease detection","authors":"Melina Maria Afonso ,&nbsp;Damodar Reddy Edla ,&nbsp;R. Ravinder Reddy ,&nbsp;Y. Ramadevi","doi":"10.1016/j.compbiomed.2025.111012","DOIUrl":"10.1016/j.compbiomed.2025.111012","url":null,"abstract":"<div><div>Parkinson’s disease (PD) detection using electroencephalography (EEG) and deep learning(DL) has garnered significant attention due to the disease’s complex nature and lack of definitive biomarkers. While imaging and invasive methods are expensive and unsafe, EEG allows for safe, low-cost, non-invasive, and portable data collection, and DL helps improve its ability to find subtle brain activity patterns. The extraction of novel features using DL yields high-dimensional data, which leads to added computational demands. This research proposes a method for feature selection that includes a novel multi-stage hybrid quorum sensing optimization (HQSO) algorithm. In the first stage, a coarse set of relevant features is selected. Quorum sensing is adopted as a performance-driven decision-making framework, where solutions adapt their behavior based on population-level fitness. A multilayer perceptron (MLP) is used to evaluate the fitness. In the second stage, a hybrid ranking mechanism statistically refines this set by prioritizing the most discriminative features, which are further reduced by a correlation-based pruning method to reduce feature redundancy. The proposed multi-stage hybrid feature selection method, integrating metaheuristic-based exploration with statistical refinement and correlation pruning, significantly improves PD detection performance. On the San Diego dataset, it achieves 98.09% accuracy, outperforming recent models like Rizvi et al. (97.90%) and matching top-tier models such as Khare et al. (100%). On the University of New Mexico dataset, it attains 94.96% accuracy, closely competing with the best-reported 99.9% by Shirisha et al. The reduction of features by almost 60% highlights its practicality for real-time, resource-constrained applications like wearable EEG-based monitoring systems.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"197 ","pages":"Article 111012"},"PeriodicalIF":6.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145044938","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}