Bioengineering最新文献

{"title":"Cross-Dataset Generalization of Deep Learning for Retinal Vessel Segmentation in Fundus Images: External Evaluation from DRIVE to CHASE_DB1.","authors":"Rashid Abbasi, Oyewale Ibrahim Abdulrasheed, Madiha Jabeen Abbasi, Md Belal Bin Heyat, Farhan Amin, Isabel de la Torre, Mirtha Silvana Garat de Marin, Eduardo Silva Alvarado","doi":"10.3390/bioengineering13101215","DOIUrl":"10.3390/bioengineering13101215","url":null,"abstract":"<p><p>Automated retinal vessel segmentation can support quantitative analysis of fundus photographs and facilitate measurement of vascular characteristics associated with ocular and systemic disease. Although deep-learning methods achieve high segmentation performance on benchmark datasets, their reliability may decrease when applied to images obtained under different acquisition conditions. This study evaluated the cross-dataset generalizability of a U-Net segmentation architecture incorporating an ImageNet-pretrained ResNet34 encoder. Sixteen images from the DRIVE dataset were used for model training and four were retained for in-domain validation. Images were resized to 256 × 256 pixels and processed using a consistent normalization pipeline, while spatial and photometric augmentation was applied during training. Without target-domain retraining or domain adaptation, the trained model was subsequently evaluated on the independent CHASE_DB1 dataset. Segmentation performance was assessed using the Dice coefficient, intersection over union (IoU), and sensitivity, together with qualitative examination of predicted vessel maps. On the held-out DRIVE images, the model achieved a Dice coefficient of 0.7263, IoU of 0.5703, and sensitivity of 0.7201. On CHASE_DB1, performance decreased to 0.6269, 0.4573, and 0.5502, respectively. The disproportionately larger reduction in sensitivity indicated an increased frequency of missed vessel pixels under cross-dataset evaluation. Qualitative examination showed greater fragmentation and omission of fine and peripheral vascular structures in CHASE_DB1 images. These findings demonstrate that within-dataset performance may overestimate the robustness of retinal vessel segmentation models when transferred to a different imaging domain. External evaluation should therefore form an important component of retinal imaging artificial intelligence research. Future work should investigate higher-resolution processing, vessel-width-specific evaluation, multi-domain training, domain-adaptation strategies, and larger multicentre datasets to improve model generalizability.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701844/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Knowledge Transfer Method from Multi-Channel to Single-Channel EEG Toward Real-World Emotion Recognition Deployment.","authors":"Yanru Gu, Hanzhen Ouyang, Zihua Ouyang, Lishengsa Yue, Huajuan Mao, Weihui Dai, Yan Kang","doi":"10.3390/bioengineering13101223","DOIUrl":"10.3390/bioengineering13101223","url":null,"abstract":"<p><p>Multi-channel electroencephalography (EEG) provides spatial information for emotion recognition but requires more complex signal acquisition than single-channel EEG. This study proposes a multi-channel-to-single-channel knowledge transfer framework that uses multi-channel recordings during training and one selected physical EEG channel at inference. A 32-channel teacher is pretrained, and Sparsemax-based gating is jointly optimized with a student using classification, logits distillation, frequency-band consistency, and gating regularization losses. The frequency-band objective compares signal-level spectral statistics of the multi-channel recordings and the gated surrogate input. After channel locking, the student is retrained on the selected physical signal. Binary Arousal and Valence classification was evaluated on DEAP using 32-fold leave-one-subject-out cross-validation. The proposed achieved higher subject-averaged balanced accuracy and F1-score than Channel-matched CE in both tasks, although component effects varied across tasks and metrics. These findings support the evaluated training approach under single-channel input constraints and provide a methodological basis for simplifying EEG acquisition.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701340/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"EETANet: A Human-Centered Intelligent Framework for Exercise Posture Quality Assessment.","authors":"Esraa Hassan, Amira Samy Talaat, Aliya Aleryani, Rana Albelaihi, M A Elsabagh","doi":"10.3390/bioengineering13101212","DOIUrl":"10.3390/bioengineering13101212","url":null,"abstract":"<p><p>As physical rehabilitation and fitness monitoring increasingly shift toward autonomous and home-based healthcare, accurate and intelligent posture assessment systems have become essential for supporting personalized rehabilitation and clinical decision making. This study proposes the Edge-Enhanced Temporal Attention Network (EETANet), a human-centered ambient intelligence framework for automated exercise posture assessment and quality evaluation. The proposed framework is designed as an end-to-end computational pipeline that transforms raw multivariate motion signals into posture quality predictions using a compact set of core biomechanical posture features. The pipeline integrates data preprocessing, signal enhancement, multi-domain feature representation, and deep neural classification to capture the relevant spatial and temporal characteristics of human movement. Signal preprocessing incorporates data cleaning, z-score normalization, and Savitzky-Golay smoothing to improve signal quality while preserving temporal dynamics. Subsequently, informative posture representations are extracted through complementary temporal, spectral, and morphological descriptors, enabling the model to characterize motion variations associated with posture quality. These representations are learned using a Multilayer Perceptron (MLP) classifier within the proposed EETANet framework to model nonlinear relationships between posture features and quality levels. Experimental evaluation demonstrates that the compact five-feature configuration achieves an accuracy of 90.83%, precision of 90.99%, recall of 90.83%, and F1-score of 90.84%. Notably, this compact configuration outperforms the corresponding 24-feature MLP configuration, which achieves 88.83% accuracy, indicating that a reduced set of core posture features can provide effective discriminative information while avoiding unnecessary feature redundancy. These findings demonstrate the potential of EETANet as an efficient framework for automated posture quality assessment in rehabilitation, fitness monitoring, and ambient healthcare applications.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701242/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261603","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ASC Deficiency Compromises Skeletal Architecture and Mechanical Competence in Mice.","authors":"Qilin Pei, Yutao He, Yun Bai, Lingxia He, Xi Li, Bin Wang, Fengning Chuan","doi":"10.3390/bioengineering13101211","DOIUrl":"10.3390/bioengineering13101211","url":null,"abstract":"<p><p>Apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) is a key adaptor of canonical inflammasome signaling and has been implicated in osteoblast differentiation and bone formation. However, whether ASC deficiency alters the structural and mechanical properties of the intact skeleton remains unclear. In this study, the skeletal phenotype of 9-week-old male <i>Asc</i>-homozygous-deficient (<i>Asc^-/-</i>) mice was characterized using micro-computed tomography, three-point bending, tibial microindentation, and silver staining of the osteocyte lacunar-canalicular system (LCS). Trabecular microarchitecture, cortical geometry, whole-bone mechanical behavior, tissue-level hardness, and LCS morphology were quantified. Associations between cortical geometric parameters and whole-bone mechanical outcomes were assessed by Pearson correlation analysis. <i>Asc^-/-</i> mice had shorter femora despite greater body weight. Trabecular bone volume fraction, thickness, and number were reduced, whereas trabecular separation was increased. Cortical analysis showed reductions in total cross-sectional area, cortical area, cortical thickness, periosteal perimeter, minimum principal moment of inertia, and polar moment of inertia. Femoral maximum load, stiffness, yield force, and work to fracture were reduced, together with lower tibial cortical hardness. ASC deficiency was also associated with enlarged osteocyte lacunae, reduced canalicular length and density, and increased canalicular width. Cortical area, thickness, and moment of inertia showed strong positive associations with whole-bone mechanical outcomes. Overall, ASC deficiency is associated with a distinct skeletal phenotype characterized by impaired trabecular architecture, reduced cortical size and load-bearing geometry, decreased whole-bone mechanical competence, lower tissue-level hardness, and altered osteocyte LCS organization. These findings extend the known skeletal role of ASC beyond osteogenesis and bone repair to the structural and mechanical phenotype of the intact skeleton.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701757/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261572","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"VLM-Based Mesh Segmentation Framework for DDH Robotic Ultrasound.","authors":"Yuanbo Wang, Jianwei Cui","doi":"10.3390/bioengineering13101213","DOIUrl":"10.3390/bioengineering13101213","url":null,"abstract":"<p><p>For robotic ultrasound targeting developmental dysplasia of the hip (DDH), reconstructed infant body-surface meshes should preserve the target surface while excluding bed, support, and background geometry. This study proposes a structured vision-language model (VLM)-based multi-view segmentation method for infant phantom mesh cleanup. Text-guided 2D masks provide foreground and background evidence; mask-quality statistics and structured VLM assessments generate view-specific foreground-support weights that undergo deterministic risk suppression before target/non-target classification of 3D vertices. Hard voting and Rule-weighted voting were baselines under identical projection and post-processing settings. We evaluated one physical infant phantom across 150 separate acquisition-reconstruction trials under controlled conditions, each comprising 40 voting views and 10 held-out evaluation views, yielding 1500 manually annotated reference masks. Structured VLM voting achieved a Balanced score (intersection over union minus the 2-pixel tolerant leakage rate) of 0.804 ± 0.085, versus 0.801 ± 0.084 and 0.793 ± 0.086 for Rule-weighted and Hard voting. This leakage rate was 0.087 ± 0.043, versus 0.089 ± 0.042 and 0.092 ± 0.043; mean Coverage remained similar (0.992-0.993). The Balanced-score gain and leakage-rate reduction relative to Rule-weighted voting remained significant after Holm correction (adjusted <i>p</i> = 0.029 and <i>p</i> = 0.033, respectively). Validation using real-infant data, 3D reference standards, and closed-loop robotic scanning is still required.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701739/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261809","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Triaxial Accelerometry Sensor for Respiratory-Effort Phase Landmark Detection During Sleep: A Cross-Modal Weak Supervision Approach.","authors":"Mengshuang Lv, Jianhong Liao, Yunhan Shi, Bowen Zhang, Qiying Zeng, Yanru Li, Demin Han","doi":"10.3390/bioengineering13101205","DOIUrl":"10.3390/bioengineering13101205","url":null,"abstract":"<p><p>Reliable respiratory phase sensing is important for respiratory-synchronized hypoglossal nerve stimulation (HGNS), but conventional sensing approaches may impose additional burden. This pilot study investigated whether a surface triaxial accelerometry sensor (TAS) could detect thoracic respiratory inductance plethysmography (RIP) troughs, which were used as operational phase landmarks representing the transition from end expiration to subsequent inspiration, rather than as directly established physiological respiratory onsets. Ten participants underwent synchronized overnight polysomnography and surface TAS recording, providing 5377.3 min of analyzable data and 75,499 automatically generated RIP landmarks. Two complementary Conformer models were trained using Gaussian weak labels and combined using multi-offset dual-conformer fusion. Generalization was evaluated using outer leave-one-subject-out validation, with fusion weights and event thresholds selected from training-side validation data. At a ±1.0 s matching tolerance, the fusion achieved participant-mean precision, recall, and F1 scores of 0.78, 0.82, and 0.80, respectively, with a 95% confidence interval of 0.75-0.85 for the mean F1 score. It outperformed each component model in 9 of 10 participants; however, its F1 gain over otherwise identical single-grid fusion was 0.0036, below the prespecified exploratory advancement margin. These findings provide preliminary evidence that surface TASs can recover an RIP-trough-derived operational phase landmark in unseen same-center participants. Independent multimodal adjudication, external replication, and full-night causal testing are required before clinical or respiratory-synchronized HGNS applications.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701435/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261672","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Time Reversal Fusion for ECG Reconstruction Before and After Adversarial Fine-Tuning.","authors":"Jiaqi Liu, Kwok Tai Chui, Lap-Kei Lee, Hao Ran Chi, Tik Yung Yip","doi":"10.3390/bioengineering13101207","DOIUrl":"10.3390/bioengineering13101207","url":null,"abstract":"<p><p>Temporal reversal preserves electrocardiogram (ECG) samples while changing their presentation to a reconstruction model. We evaluated forward/reverse (F/R) fusion against matched forward/forward (F/F) ensembles using twelve-lead, seven-second windows from the normal-labeled CODE-15% subset. Each seed block comprised two forward branches and one reverse branch, with paired initialization for the second forward and reverse branches. Reverse outputs were restored to the original time axis before fusion. Across ten seed blocks of a temporal convolutional model trained with <i>L</i>₁ loss, equal F/R fusion reduced test mean absolute error (MAE) by 1.437% relative to equal F/F, improving every block. Local adaptive F/R fusion reduced MAE by 2.140% against the same baseline. A matched continuation experiment compared <i>L</i>₁ training with least-squares adversarial fine-tuning across six seed blocks. After adversarial training, equal F/R retained a mean MAE reduction of 0.717%, improving all six blocks. Lower residual cross-moments were consistent with complementary reconstruction errors. Equal F/R also improved seed-averaged amplitude error, correlation and R-peak <i>F</i>₁ across the original-input condition and eighteen noise settings. Boundary-excluded evaluation and symmetric-padding controls retained mean directional benefits. These results support temporal direction as a source of reconstruction diversity beyond ordinary ensembling, with benefits retained under adversarial training.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701991/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial Intelligence in the Assessment and Management of Spasticity: A Scoping Review.","authors":"Ruoxi Cao, Ruojin Li, Joon Sin Ser, Yeow Leng Tan, Jing Chen","doi":"10.3390/bioengineering13101209","DOIUrl":"10.3390/bioengineering13101209","url":null,"abstract":"<p><p>Spasticity is a prevalent and disabling consequence of upper motor neuron disorders, yet its assessment and management remains clinically challenging. Artificial intelligence (AI) may offer opportunities to enhance the objective assessment, prediction, and management of spasticity; however, the current evidence base has not been comprehensively characterised. This review systematically mapped evidence on the application of AI across the continuum of spasticity care and identified key evidence gaps to inform future research. PubMed, Embase, CINAHL, and Web of Science were searched for peer-reviewed English-language studies evaluating AI in adult spasticity. Two reviewers independently screened studies, extracted data, and assessed methodological quality using NHLBI tools and PROBAST + AI. This scoping review was prospectively registered with the Open Science Framework Registries. The search yielded 1832 records, of which 21 studies were included after screening. One additional study was identified through citation searching, bringing the total to 22 studies involving 1114 participants. Applications spanned four clinical domains: discrimination between spastic and healthy individuals (<i>n</i> = 2), severity assessment (<i>n</i> = 18), risk prediction (<i>n</i> = 1), and treatment response prediction (<i>n</i> = 1). Most studies focused on post-stroke upper limb spasticity. Surface electromyography was the most frequently used input modality, often integrated with kinematic or kinetic sensors. Conventional machine learning predominated, with reported accuracies of 74-96% and the Modified Ashworth Scale as the most common reference standard. All 22 studies were at high overall risk of bias on PROBAST + AI, and none performed external validation, highlighting substantial methodological limitations that constrain clinical translation.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701998/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261625","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluation of Wearable Sensor Placement for Movement Bout Duration Estimation: A Comparative Feasibility Study.","authors":"Serena Cerfoglio, Lucia Donno, Mattia Lovato, Carlalberto Francia, Manuela Galli, Veronica Cimolin","doi":"10.3390/bioengineering13101206","DOIUrl":"10.3390/bioengineering13101206","url":null,"abstract":"<p><p>Wearable sensor placement can influence movement quantification, yet direct comparisons across anatomical sites remain limited for bout-level temporal outcomes. This study evaluated seven IMU placements for movement bout duration estimation. Twenty-seven healthy adults performed three whole-body tasks (walking, Timed Up and Go, and Step Test) and two upper-limb tasks (reaching and hand-to-mouth) under concurrent optoelectronic motion capture. IMU-derived durations were compared with segment-specific local references and, for whole-body tasks, a common sacral reference used as a proxy for whole-body movement. Agreement was assessed using Spearman's correlation, ICC, Bland-Altman analysis, and MAPE. Participant-level agreement analyses were complemented by trial-level repeated-measures analyses to account for within-participant variability across repeated trials. A descriptive composite ranking combined ICC, MAPE, and limits-of-agreement width against the common reference. Performance varied across placements and tasks, with the largest local-common dissociation observed during the TUG. The sacrum obtained the most favorable overall composite ranking score (2.78), closely followed by the dominant ankle (2.89) and non-dominant wrist (3.11). Both wrists showed high agreement during upper-limb tasks (ICC = 0.879-0.994). Trial-level analyses confirmed a significant effect of sensor placement for all whole-body tasks, while no significant differences were observed between wrist placements during the isolated upper-limb tasks. The non-dominant wrist showed a favorable combination of performance across movement domains, supporting its selection as a promising candidate for further evaluation in subsequent free-living monitoring.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701408/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Physiology-Informed Blink Artefact Detection from Bilateral Frontal EEG Under Asymmetric Interference.","authors":"Weiqiang Zhang, Menglong Li, Yujie Zhang, Jingyi Cao, Feng Shen, Xiaohui Liu, Chen Chen, Yang Li","doi":"10.3390/bioengineering13101208","DOIUrl":"10.3390/bioengineering13101208","url":null,"abstract":"<p><p>Eye blinks contaminate frontal electroencephalography (EEG), while asymmetric disturbances can cause false detections. We present a physiology-informed detector that combines waveform correlation, peak-time agreement, amplitude similarity, and symmetric-component dominance in an explicit event-verification rule. It uses two frontal channels without supervised model fitting. Evaluation on raw ZuCo 1.0 recordings included 260 eligible blocks from 12 participants, totalling 1291.61 min and 8611 eye-tracker reference intervals. Scalp and periocular pairs were evaluated separately. On the scalp pair, participant-mean precision was 0.731, recall 0.774, F1 0.734, and false positives 1.469 per minute. At fixed settings, false-positive rates were lower than with single-channel thresholding and bilateral averaging (both Holm-adjusted <i>p</i> = 0.005859), and remained lower than those of four rule-based controls across five perturbation conditions. Threshold-selected simple controls achieved higher precision at similar recall. Component ablations quantified the effects of the verification checks on sensitivity and false positives. The offline Python detector required a median of 2.13 ms per minute of recorded data on an Apple M1 Pro. These results establish a computationally inexpensive approach that makes blink-event decisions directly inspectable through bilateral signal measurements.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":"13 10","pages":""},"PeriodicalIF":4.4,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13701908/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149261652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}