Bioengineering

In vitro fertilization (IVF) laboratories equipped with timelapse incubators generate vast quantities of sequential embryo images, yet the absence of standardized, annotated databases impedes the development of reproducible computational tools for embryo assessment. Here we describe the construction of a standardized time-lapse imaging database comprising 631 normally fertilized zygotes from 218 treatment cycles, integrating timelapse image sequences, patient clinical records, and embryo developmental outcomes. We further present a gradient boosting decision tree (GBDT) ensemble framework that integrates zygote morphokinetic parameters-continuous time-series features extracted via a validated CNN-based segmentation algorithm (US Patent US11210494B2)-with conventional embryo assessment grades (categorical features per the Istanbul consensus). The fusion framework employs equal-weight initialization followed by iterative residual-decreasing training to optimally combine heterogeneous feature types. Ablation analysis demonstrated that the integrated model achieved an AUC of 0.78, significantly outperforming morphokinetics-only (AUC 0.71) and conventional-only (AUC 0.65) models, confirming the complementary value of the two data modalities. The database and fusion framework provide a reproducible foundation for embryo development assessment and are generalizable to other multimodal data integration tasks in reproductive medicine.

Ovarian cancer is recognized as the deadliest gynecological malignancy. Diagnosis at advanced stages and the lack of effective screening program lead to poor survival rates, dropping to 17-39 % in stage III-IV diseases. Ultrasound (US) is the primary imaging modality for ovarian structures evaluation, but it is strongly affected by the operator expertise due to the complexity of adnexal masses and the physiological variability of ovarian morphology throughout a womans lifecycle. The International Ovarian Tumor Analysis (IOTA) group introduced definitions and predictive tools to standardize gynecological US interpretation. However, these tools still rely on subjective interpretation, thus highlighting the need for more objective solutions. Recent studies have explored artificial intelligence (AI) algorithms for gynecological US, mainly focusing on adnexal masses classification. Conversely, a robust solution supporting the identification and description of healthy and tumoral ovarian structures is still lacking. This paper proposes OvAi Focus, a framework including (i) a segmentation module for the identification of healthy ovaries, plus solid and cystic components of adnexal masses; (ii) a morphology module for the extraction of IOTA-based keywords to describe adnexal masses morphology. Segmentation module results were compared to ground truth masks, showing DICE scores from 0.62 for functional ovary to 0.87 for the whole adnexal mass. Morphology module was tested through interobserver agreement analysis, obtaining Fleiss Kappa from 0.16 to 0.57 and Percent Agreement from 47 to 90 %, in line with existing literature. OvAi Focus represents an innovative solution which could help overcoming subjectivity in gynecological US imaging interpretation.

Background: Data-extraction forms are tested before use in systematic reviews, but the guidance and error studies behind that practice concern numerical data. When the datum is a term transcribed verbatim from a source, as in evidence maps that inventory terminology, neither an agreement measure nor extraction conventions are established. We piloted, before Phase 1 of the ATLAS-ONTO project, a form built to extract operative-step terms from the surgical literature on deep endometriosis, to determine whether it produced acceptable inter-reviewer agreement and what revisions it required. Methods: Blinded inter-reviewer agreement study, reported according to GRRAS. Two reviewers independently extracted terms from 11 purposively heterogeneous articles in English and French. The primary measure was the mean per-article Jaccard index of normalized term sets, with an explicit rule for empty sets; secondary measures were Cohen's kappa for categorical attributes of exactly matching terms and simple agreement for anatomical structure. Thresholds were fixed before extraction (Jaccard >=0.70; kappa >=0.60; simple agreement >=0.80). Unmatched terms were classified retrospectively by mechanism, and the same articles were reassessed after a harmonization session. Results: Eight articles were extracted, seven with a defined index. The form failed its threshold: the mean Jaccard index was 0.581 (pooled 55/92 = 0.598). Of 92 terms, 55 matched exactly and 37 did not; the unmatched terms were attributed to source coverage (21), span extent (8), source-language comprehension (7) and one residual discrepancy. kappa conditional on the 55 matching terms was high (end definition 0.930; laterality 0.781) and anatomical-structure agreement was 0.873, so kappa alone did not detect the failure. Two record-integrity defects found during analysis are reported with their effect. Four written conventions and a split start-definition field followed; reassessment after harmonization reached 0.913, which is convergence, not independent reproducibility. Conclusions: A form that extracts terminology or verbatim text should be piloted with a set-overlap measure, a rule for empty sets and a defined admissible source, against a threshold fixed in advance. The mechanisms of discordance found here are properties of text extraction and are not specific to surgery. The revised form will be tested on new material in Phase 1.

Objective: To perform internal and external validation of a gradient-boosted decision tree (GBDT) fusion model that integrates zygote morphokinetic parameters with conventional embryo assessment features for blastocyst prediction, and to compare its discriminative performance against senior embryologists. Methods: This retrospective cohort study included 631 normally fertilized zygotes from 218 treatment cycles. A GBDT fusion model integrating 84 zygote morphokinetic parameters and 8 conventional assessment features was evaluated internally (5-fold cross-validation) and externally on a public dataset of 523 embryos with blastocyst outcomes. Model performance was assessed using area under the ROC curve (AUC), area under the precision-recall curve (AUPRC), F1 score, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Discrimination was compared with embryologist consensus using the DeLong test; agreement was assessed with Cohen's kappa. Results: The model achieved an internal AUC of 0.78 (95% CI 0.74-0.82), AUPRC 0.72, F1 0.73, sensitivity 0.74, specificity 0.77, PPV 0.72, and NPV 0.79. External validation on the public dataset demonstrated acceptable generalizability (AUC 0.76, 95% CI 0.71-0.81). The model significantly outperformed embryologist consensus (AUC 0.70, P<0.001) with moderate agreement (kappa=0.56). Decision curve analysis confirmed clinical net benefit at threshold probabilities of 0.15-0.55. Conclusions: The GBDT fusion model integrating zygote morphokinetics with conventional assessment demonstrates good discrimination and external generalizability for blastocyst prediction, providing an interpretable decision-support tool for embryo selection in IVF practice.

Musculoskeletal simulations are widely used to estimate joint loading, yet most musculoskeletal models estimate knee contact forces as resultant forces or as normal medial and lateral contact forces, without resolving pressure distributions across the articular surfaces. This paper presents a smoothed mesh-based knee contact pressure model that computes continuously differentiable tibiofemoral contact pressures, enabling its direct integration into full-body movement simulations. Built on an elastic foundation formulation, the model introduces smooth approximations ensuring that all contact functions and their derivatives remain continuous throughout the simulation. A systematic sensitivity analysis was performed across five key parameters: mesh resolution, joint damping and three smoothing parameters. Tracking simulations across eight gait trials demonstrated that the nominal configuration achieved mean RMSE values for medial and lateral knee contact forces of 51.6 N and 75.2 N, respectively, with a mean RMSE for joint angles of 1.45{degrees} and (r = 0.97), converging in less than three hours on a standard computer. Mesh resolution was identified as the dominant factor that affected both accuracy and convergence, while damping variations had negligible influence. As a proof of concept, the model was also incorporated into predictive simulations, demonstrating that increasing the weight on the contact pressure term in the cost function leads to reduced tibiofemoral loading, particularly in the lateral compartment. The proposed formulation provides a computationally efficient and numerically robust framework for simulating knee contact mechanics within full-body musculoskeletal models.

Therapeutic footwear with rocker soles is widely used to reduce plantar pressure in individuals at risk of diabetic foot ulceration. Of the various rocker sole characteristics, apex position, rocker angle, and apex angle have been extensively studied; however, the effect of forefoot rocker curvature has been seldom investigated. The aim of this study was to assess the isolated effect of forefoot rocker curvature on plantar pressure distribution, lower limb gait kinematics, and comfort. Ten healthy adults walked in a modular footwear setup with three interchangeable midsoles featuring a straight-line rocker profile (CRi), a rocker radius of 60% (CR60), and 45% (CR45) of the shoe length. In-shoe plantar pressure was measured using 99-sensor capacitive insoles. Kinematics of lower limb joints were measured via inertial measurement units. Comfort-related parameters were assessed using a Visual Analog Scale. Data were analyzed using non- parametric tests. CRi demonstrated significantly lower pressure at the metatarsal heads than CR60 and CR45 (kPa; median[Q1-Q3]; mean pressure=80 [67-81]; peak pressure=151 [139-169]; p=0.002). CR45 showed significantly lower mean pressure and pressure-time integral in the midfoot than CRi (p=0.002). CR60 showed the largest maximum dorsiflexion angle (deg, median [Q1-Q3]: CR60=17.8[13.4-23.5]; p<0.001) and range of motion (deg: CR60=20.3[18.8-23.3]; p=0.007). CRi was perceived as the least comfortable rocker condition (score 0-100; median[Q1-Q3]; overall comfort=44 [35-57]; p=0.008). A straight rocker maximizes offloading at the expense of ankle kinematics and comfort; lower rocker radii enhance comfort, thus making the rocker curvature a key design parameter for therapeutic footwear.

Predicting implantation outcomes in vitro fertilization (IVF) is challenging because clinically relevant information becomes available at different points during treatment. In this study, we developed a stage-aware, leakage-conscious, and explainable machine-learning framework to predict implantation using a public IVF dataset. {beta}-HCG was excluded to avoid potential downstream target leakage. The data were divided into training, validation, and independent test sets. Several machine-learning models, including Logistic Regression, Random Forest (RF), Extra Trees, XGBoost, and Hist Gradient Boosting, were optimized using stratified cross-validation. A fixed RF model was then evaluated at three stages of the IVF process, using baseline couple characteristics, gamete and fertilization variables, and embryological variables, respectively. Model performance improved as additional treatment-related information became available, with the largest improvement observed after the inclusion of embryological variables. The final RF model showed good predictive performance on the independent test set. The number of embryos was consistently identified as the most important predictor across RF, permutation, and SHAP analyses. Overall, the findings suggest that IVF implantation prediction becomes more accurate as later-stage treatment information is incorporated, particularly information related to embryological characteristics. The proposed approach also provides interpretable predictions while reducing the risk of information leakage. Further evaluation using independent clinical datasets is needed before the framework can be considered for clinical use.

Obstetric anal sphincter injury (OASI) requires immediate postpartum recognition for primary repair. Diagnosis relies primarily on clinical examination, as endoanal ultrasound (EAUS) is rarely a bedside option. Therefore, objective diagnostic assessment is most needed during the first postpartum hours. However, it is challenging to obtain the anatomically verified data needed to develop diagnostic technologies for this clinical window. We evaluated the translation of a machine-learning-assisted impedance spectroscopy model into its intended immediate postpartum setting. The model lifecycle comprised three sequential phases: V1 development using anatomically verified data from a prospective multicenter study (N=152); V2 real-world adaptation, comprising post-market data collection, retraining (N=144) to adapt to acute physiological tissue changes, and subsequent separate evaluation (N=125); and prospective external validation of V2 in an unseen cohort (N=70). During this separate evaluation, V2 achieved 89.6% accuracy. In external validation, all examinations occurred within 4 hours of birth (94.3% within the first hour). Against subsequent EAUS, V2 achieved 86.2% sensitivity, 92.7% specificity, and 90.0% overall accuracy. The V2 model was prospectively externally validated in its intended immediate-postpartum setting, supporting ML-assisted impedance spectroscopy as an objective adjunct, and not a replacement, to clinical OASI assessment. Further validation is required to confirm generalizability.

Successful instrumented fusion of the lumbar spine is a complex surgical challenge, with positive patient outcomes dependent on careful surgical planning. Material selection is of critical importance to a mechanical construct supporting successful spinal fusion. Therefore, the aims of this study were to (a) evaluate the potential clinical use of finite element analysis (FEA) and (b) conduct a retrospective mechanical analysis of different implant materials in patients having undergone spinal fusion using FEA. Our methodology involved segmenting the spine from post-operative computed tomography (CT) image data from patients with previous spinal fusion. FEA models representing post-surgery cases were developed and different biomechanical loading conditions such as compression, flexion, bending and extension whilst testing pedicle screws of different materials were simulated. Patient specific finite element models were created, and biomechanical analysis were completed for all three patients. Polyetheretherketone (PEEK) constructs typically demonstrated lower peak implant stress when compared to titanium constructs for all spinal fusion levels. Furthermore, increasing the spinal fusion level resulted in significant differences in the maximum von Mises stress within both the bone and the instrumentation, whereas the 2-level fusion exhibited comparable stress levels in the bone irrespective of the instrumentation material. This pilot explores the potential of FEA as a clinical tool for assessing device and bone stresses. In our cohort, different materials can influence the stresses in both the instrumentation and the instrumented vertebrae, suggesting FEA can be useful pre- operative tool with regards to instrument selection and post-operatively to assess instrumentation and bone stresses. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=74 SRC="FIGDIR/small/742715v1_ufig1.gif" ALT="Figure 1"> View larger version (33K): org.highwire.dtl.DTLVardef@55959aorg.highwire.dtl.DTLVardef@d0b9d6org.highwire.dtl.DTLVardef@158c348org.highwire.dtl.DTLVardef@7ce828_HPS_FORMAT_FIGEXP M_FIG C_FIG

Predicting wear in artificial joints requires integrating joint dynamics, contact mechanics and progressive surface evolution, yet these processes are often treated separately. In total hip arthroplasty (THA), finite-element approaches remain the reference standard, but they are computationally demanding and usually rely on boundary conditions from independent musculoskeletal (MSK) models, hindering consistent coupling and feedback between wear progression and movement dynamics. As a single-subject proof of concept, we present a computational framework that embeds wear estimation within forward MSK simulations through OpenSim-MATLAB integration. Contact variables are computed using an elastic-foundation formulation, and wear is updated through the Archard law, enabling cyclic prediction of contact mechanics and surface evolution within a single workflow at practical computational cost. The framework was evaluated in one subject with right THA during five activities of daily living and numerically benchmarked against finite-element simulations. A long-term walking analysis of 4 million cycles was also performed to assess geometry updating. Across tasks, peak contact pressures remained within 7% of finite-element predictions. Linear wear depth and volumetric loss showed maximum deviations of 16% and 13%, respectively. Accounting for progressive geometry changes yielded a maximum wear depth about 31% lower than linear extrapolation. These preliminary results support the framework's computational feasibility and numerical consistency for the tested case; nevertheless, multi-subject evaluation is required before broader predictive or clinical use.

Automated segmentation of Optical Coherence Tomography (OCT) images is a critical component of structural biomarker extraction for retinal diagnostics. Deep learning models achieve state-of-the-art performance on controlled datasets, yet exhibit unpredictable failures on real-world data. Current quality gates rely on device-reported scan quality scores, which have been shown to be unreliable predictors of segmentation performance. We define scan quality in a task-specific sense, that is, whether a given B-scan will yield a reliable segmentation from a particular trained model. Under this definition, we systematically evaluate No-Reference Image Quality Assessment (NR-IQA) metrics, general-purpose pretrained representations, and domain-specific pretrained representations as alternative quality gates. We use choroid segmentation as the prototype task, with a dataset of 6,076 OCT B-scans from 80 subjects. These quality gate candidates are evaluated at three levels: scalar metrics (BRISQUE, NIQE, PIQE, SNR, PSNR), supervised linear probing, and unsupervised partitioning (K-Means) of feature vectors and learned representations. All scalar NR-IQA metrics proved inadequate (|r| < 0.20). General-purpose pretrained representations (EfficientNet-B0, ResNet-50, ViT-B/16) outperform NR-IQA, achieving ROC-AUC up to 0.77, indicating that learned representations are better suited to task-specific quality gating than hand-crafted scalar statistics. Retinal foundation models (FMs) further improve performance: RETFound, an OCT-specific FM, achieves ROC-AUC of approximately 0.81. Unsupervised K-Means partitioning of the embeddings recovers the same hierarchy geometrically: only the two retinal foundation models produce quality-aligned clusters that exceed a patient-level permutation null, while NR-IQA and general-purpose pretrained feature spaces do not, suggesting that domain-specific pretraining provides an additional benefit beyond general-purpose learned representations.

Background: Pain relief is the principal patient-centered goal of surgery for symptomatic lumbar synovial facet cysts, yet comparative reviews have often emphasized cyst recurrence. Whether adding fusion improves postoperative pain or reduces later surgery remains uncertain. Objective: To compare decompression alone with decompression plus fusion, with postoperative back- and leg-pain outcomes as the primary domain. Methods: PubMed, Embase, Scopus, Web of Science, and the Cochrane Library were searched from inception to 2 June 2026. Comparative cohorts and case series with at least five patients were eligible. Twenty-two studies were re-extracted for VAS/NRS scores, change scores, and persistent or recurrent pain. Random-effects restricted maximum likelihood models with Hartung-Knapp inference were used; clinically distinct pain outcomes were analyzed separately. Results: Twenty-two studies (16 cohorts, 6 case series; 51,899 participants) were included. Two studies provided compatible final VAS data. Fusion did not improve postoperative back pain (MD -0.04, 95% CI -0.17 to 0.10; I2=0%) or leg pain (MD -0.03, 95% CI -0.28 to 0.21; I2=0%). Postoperative back pain (RR 0.58, 95% CI 0.14-2.30) and leg/radicular symptoms (RR 0.75, 95% CI 0.42-1.32) were also not significantly reduced. Fusion decreased confirmed cyst recurrence (RR 0.29, 95% CI 0.15-0.57) but not reoperation or subsequent lumbar surgery (RR 0.80, 95% CI 0.42-1.50). Conclusion: Current comparative evidence does not demonstrate superior postoperative pain control with routine fusion. Fusion reduces cyst recurrence without clearly reducing reoperation, supporting selective use when instability is present or anticipated.

This study aims to improve the detection performance of age-related macular degeneration (AMD) in low-quality retinal images. BackgroundAMD is a leading cause of vision loss among older adults globally, and accurate detection is crucial for clinical management. However, low-quality optical coherence tomography (OCT) images significantly compromise diagnostic accuracy. ObjectiveTo enhance AMD detection in low-quality images using noise-augmented data augmentation and an improved YOLO deep learning model. MethodsPublic datasets from UCSD and Duke University were utilized; the training dataset comprised 24,980 OCT images (high-quality and noise-augmented low-quality), while the testing dataset included 1,000 images (584 AMD, 416 normal). The model is based on the YOLOv8n framework, integrated with Squeeze-and-Excitation blocks (SEblock) and Adaptive Sparse Self-Attention (ASSA), with an additional 160x160 detection layer for detecting small lesions. Evaluation metrics included accuracy, sensitivity, specificity, and F2-score. ResultsThe proposed model achieved an accuracy of 99.02%, sensitivity of 98.17%, specificity of 100%, and an F2-score of 98.50% on the Duke dataset. Detection rates were significantly improved compared to traditional methods, particularly in low-quality images, with a detection rate of 89.60%, markedly superior to original YOLOv8n (55.10%) and classical models like ResNet50. ConclusionThe enhanced model, employing noise-augmented training data and improved attention mechanisms, demonstrates excellent AMD detection capabilities in low-quality OCT images, showing broad potential for clinical applications.

Introduction Autologous and allogeneic chondrocyte cell sheet transplantation has shown safety and potential efficacy for cartilage defects associated with knee osteoarthritis. Although polydactyly-derived chondrocyte (PDC) cell sheets have been used clinically for allogeneic transplantation, a scalable and reliable alternative cell source is needed. This study aimed to establish a proof-of-concept protocol for generating chondrocyte cell sheets from clinical-grade human induced pluripotent stem cells (iPS cells). Methods Cartilaginous tissue was generated from human iPS cells, and isolated chondrocytes were used to fabricate iPS cell-derived chondrocyte sheets. In Phase 1, sheets were produced in high-serum medium under normoxic, hypoxic, or hypoxic/high-seeding-density conditions and compared with PDC sheets using a three-dimensional in vitro model and xenogeneic orthotopic transplantation in an immunodeficient rat model. In Phase 2, sheets were produced in low-serum medium under hypoxic/high-seeding-density conditions and evaluated for cartilage-related gene expression, secreted factors, cell surface marker expression, and in vivo cartilage repair. Results In Phase 1, hypoxic/high-seeding-density culture enhanced in vitro chondrogenic properties. However, after transplantation, iPS cell-derived chondrocyte sheets formed fibrotic repair tissue, whereas PDC sheets generated hyaline-like cartilage. In Phase 2, low-serum culture increased expression of cartilage-related genes, including collagen type II alpha 1 (COL2A1) and SRY-box transcription factor 9 (SOX9), enhanced secretion of transforming growth factor beta 1 and melanoma inhibitory activity, and upregulated CD9, CD26, CD56, CD201, CD227, and GD2. Low-serum iPS cell-derived chondrocyte sheets regenerated hyaline cartilage at 4 weeks, and the repair tissue was maintained at 12 weeks without bone formation. Conclusions A robust fabrication protocol enabled human iPS cell-derived chondrocyte sheets to regenerate hyaline-like cartilage in vivo. These findings support their potential as a scalable cell source for clinical cartilage repair.

Ultrasound is commonly used to assess structural changes in symptomatic Achilles tendons, but quantitative biomechanical metrics for progressive tendon deterioration remain limited. The goal of this study was to develop and validate an automated ultrasound tracking algorithm for regional tendon deformation and evaluate strain progression in survived and ruptured tendons during fatigue loading. We hypothesized that maximum strain, average strain, and strain heterogeneity would exhibit different trajectories between groups. Ten cadaveric Achilles tendons underwent cyclic loading with stress tests every 500 cycles until rupture or 150,000 cycles. Ultrasound images acquired during stress tests were analyzed using an automated tracking algorithm to generate spatially resolved regional strain fields. Ultrasound-derived bulk strain was highly correlated with actuator-derived strain in survived (R^2 = 0.968 +/- 0.017) and ruptured tendons (R^2 = 0.972 +/- 0.014). Maximum and average longitudinal strains progressively diverged between groups across fatigue life (Group x FatigueLife: p = 0.003 and p < 0.0001, respectively). During the first 10,000 cycles, average strain decreased in survived tendons ({beta} = -0.0268%, p = 0.0215) but not ruptured tendons ({beta} = 0.0147%, p = 0.1197), with a significant Group x Cycle interaction (p = 0.0061). This study demonstrates that the algorithm quantified Achilles tendon deformation with high fidelity and enabled spatially resolved strain assessment throughout fatigue loading. Maximum and average strain followed different trajectories between groups, whereas strain heterogeneity did not. Early differences in tendon biomechanics suggest that regional strain behavior may change before pronounced differences in absolute magnitude develop.

Published artificial intelligence for keratoconus is unproven in fellow eyes that topography and tomography call normal, the eyes where surgical decisions are made, although it reports near-perfect accuracy for early disease. In a registered systematic review of 573 studies (PROSPERO CRD420261441197), none of 2,035 claim-by-stratum units from 212 adjudicated studies established discrimination in these eyes. Most units (90.0%) answered a different question, never constructed the target population or certified its normality; external validation did not test label generation, and five systematic reviews did not assess it. In a public cohort, a frozen score yielded an area under the receiver operating characteristic curve of 0.996 against a same-eye index label and 0.549 against a contralateral proxy; its inputs reconstructed the label-generating index (r = 0.991), so accuracy there measured agreement with the index. Anchor-gap evaluation measures accuracy in these eyes against the contralateral clinical diagnosis.

Background: Vertebroplasty and balloon kyphoplasty are used for symptomatic vertebral hemangiomas, although comparative evidence is limited. We summarized pain relief, cement leakage, and recurrence after vertebral augmentation and assessed whether direct comparison of the two techniques was feasible. Methods: Five databases were searched from inception to January 2, 2026, with an update on July 5, 2026. Because only one small cohort directly compared vertebroplasty with kyphoplasty, outcomes were pooled as single-arm proportions or, for early pain change, as a mean difference using random-effects models. Prespecified subgroup, sensitivity, small-study effect, and influence analyses were performed. Results: Forty-four studies were included: 33 case series, 10 cohort studies, and one randomized trial. Kyphoplasty-specific evidence comprised one dedicated series and one comparative cohort. Any cement leakage occurred in 10.5% of patients (14 studies; 95% CI 5.7-18.4%), while trim-and-fill gave an exploratory adjusted estimate of 20.4%. Early pain reduction averaged 5.13 points on a 0-10 scale (8 studies; 95% CI 4.48-5.77; I2=89.4%). Complete or near-complete pain relief occurred in 79.4% of patients (10 studies), and recurrence, progression, or retreatment occurred in 3.9% (13 studies). Symptomatic cement leakage was uncommon at 0.4%. Conclusion: The available literature, which is mainly retrospective and vertebroplasty-based, supports substantial pain relief with infrequent symptomatic complications. Kyphoplasty data remain insufficient for a reliable technique comparison. Prospective studies with standardized clinical and imaging outcomes are needed.

Background: Artificial intelligence (AI) systems for glaucoma diagnosis and prognostication from visual fields (VF) are under active development, yet do not audit for vertical-meridian-respecting field loss - known sequelae of stroke, hemorrhage, and neoplasm. We developed a self-supervised encoder of automated perimetry that learns anatomically interpretable VF structure without labels, and evaluated its capacity to identify suspected neurologic VF patterns in an independent public glaucoma dataset. Methods: We pretrained a 128-dimensional masked autoencoder on 23,223 unlabeled Humphrey VFs (patient-grouped training split of 28,943 fields from 3,871 patients; UWHVF, all-comers perimetry), using monocular pattern-deviation input. A supervised linear classifier over vertical-midline latent dimensions was trained on per-eye expert neurological/non-neurological labels and assessed under hard-negative cross-validation, with specificity evaluated on 100 held-out, structurally separated UWHVF controls. External evaluation used the Harvard-Glaucoma Fairness dataset (Harvard-GF; 3,300 patients with paired VF and optical coherence tomography [OCT] from a single academic center), which contributed no data at any training stage. Results: Masked reconstruction recovered structure concordant with retinal neuroanatomy: 50 of 128 latent dimensions emerged spatially specialized, versus 23 for the total-deviation encoder. The classifier achieved cross-validated balanced accuracy 0.78 (95% CI, 0.75-0.82) and AUC 0.85 (95% CI, 0.82-0.89), with no false positives among the 100 held-out controls. Applied to Harvard-GF without fine-tuning, it identified a top-20 of 1,748 glaucoma-labeled patients (1.1%) with morphology inconsistent with glaucoma; all 20 were positive on the rule-based Neurological Hemifield Test (mean score 62.4), and OCT showed preserved superior (Cohen d = +0.68; P < .001) and inferior (d = +0.63; P = .003) retinal nerve fiber layer versus severity-matched controls. Conclusions: A self-supervised VF encoder learned anatomically interpretable visual field structure from unlabeled data and identified suspected neurological cases in a curated glaucoma dataset, with expert, rule-based, and OCT corroboration. Visual field datasets used to train glaucoma AI may benefit from neurological screening before model training; the encoder reported here supports such audits and provides a foundation for neuro-ophthalmic AI beyond fundus photography and OCT.

Conventional transfemoral amputation disrupts native neuromuscular pathways, limiting prosthetic joint control, sensory feedback, and the perception of the prosthesis as part of the body. To ameliorate these pathologies, we restored the agonist-antagonist relationship of residual muscles in two individuals with above-knee amputation through an interventional surgical revision. Participants trained with a bionic knee prosthesis before and after the surgical revision while generating neuromuscular, cortical, functional, and affective data. Both individuals demonstrated improvements after the revision that could not readily be attributed to training effects, including: 1) increased proprioceptive afferents and stronger activation in cortical regions associated with sensorimotor integration of their missing joints, 2) improved control of the bionic knee during functional tasks including sit-to-stand and stair ascent, and 3) generally greater prosthesis embodiment, proprioception, and phantom limb definition as assessed through questionnaires and interviews. In contrast, training outcomes were more participant-specific and more variably correlated with amount of exposure, especially before the revision. These pilot findings suggest that revisional augmentation of residual neuromuscular tissues to restore agonist-antagonist dynamics may promote sensorimotor coherence and enhance both functional and perceptual integration with a bionic prosthesis, and remaining participant-specific heterogeneities may be attributable to inter-individual difference in residual limbs neuromuscular system, amputation history, and personal beliefs about prosthesis usage.

Conventional clinical assessment of upper limb kinematics relies on goniometry or visual estimation, methods that provide limited information on multi-segment 3D kinematics and can have poor inter-rater reliability. Inertial measurement units (IMUs) represent a viable alternative; however, extracting anatomically meaningful joint angles from skin-mounted sensors remains challenging owing to sensor-to-segment misalignment, soft-tissue artefact, and suboptimal orientation estimation under dynamic conditions. This study developed and validated a multi-segment IMU-based framework for upper limb kinematic assessment, addressing three stages of the measurement pipeline: (i) sensor fusion gain optimisation as a function of task mean segment angular velocity; (ii) accuracy of an anatomical landmark registration-based alignment method; and (iii) comparative evaluation of landmark-based alignment against manual and neutral-pose alignment approaches. Twenty participants with typical shoulder function performed clinical and functional upper limb tasks, recorded simultaneously with an optoelectronic motion capture system and six IMUs tracking the thorax, clavicle, scapula, humerus, forearm, and hand. Orientation estimation used a gradient-descent complementary filter with velocity-dependent gain modulation. Sensor-to-segment alignment was achieved via manual, neutral-pose, and landmark registration methods. Mean sensor orientation errors ranged from 1.5 to 3 degrees across segments/tasks and increased with angular velocity. Landmark registration produced substantially lower joint angle errors than manual and neutral-pose alignment across all joints (root mean square error range: 3.7-7.8 degrees; manual: 8.3-24.2 degrees; neutral: 15.5-35.5 degrees). These results indicate that IMU-derived kinematic accuracy is context-dependent, varying with joint and movement type, suggesting that accuracy should be established for the specific movements and joints relevant to each clinical application.