Background: Psychological readiness, quantified by the Anterior Cruciate Ligament Return to Sport after Injury (ACL-RSI) scale, is one of the most consistent correlates of return-to-sport success following ACL reconstruction. Whether psychological readiness reflects an independent construct or is related to measured functional performance is not known. Purpose: To evaluate the associations between patient-perceived knee function and objective performance on a functional task, as quantified by traditional scoring, kinematics, and motion quality, and psychological readiness 9 to 12 months after ACL reconstruction. Methods: In this cross-sectional study, forty patients performed the Star Excursion Balance Test using a validated markerless motion capture system. Three biomechanical domains derived from that task (reach distance, lower-extremity kinematics, and a time-series motion quality index) were evaluated against psychological readiness, with patient-reported function (as measured by the Knee Injury and Osteoarthritis Outcomes Score [KOOS-12]) as a comparator. Associations were estimated using multivariable regression adjusted for age, sex, body mass index, and limb length. Relative contributions were evaluated with variance decomposition, and Bayes factors quantified the strength of evidence for or against each association. Results: Patient-perceived knee function was strongly associated with psychological readiness (standardized β = 0.58; 95% confidence interval, 0.29 to 0.87; BF10 = 249), whereas no biomechanical domain derived from the functional task reached statistical significance (all |β| ≤ 0.34; all |r| ≤ 0.3). In the combined model (total R2 = 0.463), variance decomposition attributed 89.3% of explained variance in psychological readiness to patient-perceived knee function, versus 6.9% for reach distances, 2.4% for motion quality, and 1.4% for lower extremity kinematics. Bayes factors favored the null for each biomechanical domain, but reached only anecdotal strength (BF01, 1.3–2.8). Conclusion: In this cross-section, psychological readiness after ACL reconstruction more closely aligned with patient-perceived knee function than biomechanical performance on the SEBT, indicating that perceived function and biomechanical performance on certain tasks may be divergent at the time of return-to-sport clearance. These findings support the routine inclusion of patient-reported outcome measures alongside functional testing in multi-domain return-to-sport assessment.
T1ρ mapping is emerging as a potential, contrast-free alternative to late gadolinium enhancement (LGE) for assessment of myocardial viability. However, strong signal contributions from the blood pool can impede quantitative evaluation at the (sub)endocardium. In this work, we study the effectiveness of dark-blood (DB) contrast in adiabatic T1ρ (T1ρ,adiab) mapping at 3T, using slice-selective and non-selective adiabatic spin-lock pulses. Adiabatic DB-T1ρ,adiab preparations consisted of an odd number of slice-selective adiabatic full passage (AFP) pulses, followed by a final, non-selective AFP pulse. This preparation induces T1ρ,adiab decay within the imaging slice while inverting the magnetization outside. A delay (δ) between preparation and imaging allowed for relaxation and inflow of the inverted blood to achieve DB contrast. Bias and precision of DB- and bright-blood (BB)-T1ρ,adiab were compared in phantom and in healthy subjects (n = 10). Blood suppression efficacy and apparent myocardial thickness in DB imaging were investigated in simulations, phantom, and in vivo. The clinical feasibility of DB-T1ρ,adiab mapping was evaluated in a small cohort of patients (n = 7) with suspected cardiovascular diseases. DB-T1ρ,adiab values were in agreement with reference BB-T1ρ,adiab values in phantom (myocardium-like vial BB: 219.27 ± 4.80 ms, DB: 218.09 ± 8.22 ms) and in healthy subjects (BB: 182.32 ± 28.27 ms, DB: 183.49 ± 45.54 ms). A moderate increase in intra-(wCVi,r) and inter-scan variability (wCVi¯) was observed in the DB method, compared with conventional BB imaging, for phantom and healthy subjects (in vivo wCVi,r BB: 15.51 ± 2.65%, DB: 24.82 ± 4.18%; in vivo wCVi¯ BB: 3.38 ± 0.86%, DB: 7.24 ± 2.55%). Longer delay times improved blood suppression in vivo for DB-T1ρ,adiab, albeit at increased intra-scan variability in phantom and in vivo (DB wCVi,r for δ = 0 ms: 4.90 ± 0.83% in phantom, 13.44 ± 2.91% in vivo, for δ = 600 ms: 8.14 ± 1.88% in phantom, 26.25 ± 5.19% in vivo). Average apparent myocardial thickness was slightly higher when using DB-T1ρ,adiab compared with BB-T1ρ,adiab (BB: 7.33 ± 2.05 mm, DB: 7.99 ± 2.46 mm). DB-T1ρ,adiab maps yielded comparable image quality to BB-T1ρ,adiab maps in patients. DB-T1ρ,adiab mapping represents an alternative to BB-T1ρ,adiab for myocardial assessment with the potential for improved visualization of the (sub-)endocardium.
Forward head posture (FHP) is common in young adults using technological devices and often accompanied by increased thoracic kyphosis. This study reports the design and custom manufacturing of a 3D-printed craniocervical orthosis and a pilot application evaluating its effects on the craniocervical angle (CVA), thoracic kyphosis angle (TKA), and user satisfaction. In this prospective, single-arm, within-subject pilot study, each participant underwent 3D scanning, and a custom orthosis was manufactured from their scan data using a common three-point CAD template via FDM 3D printing. Ten volunteers aged 18–25 years completed the study without a priori sample size calculation. The CVA and TKA were assessed photogrammetrically at baseline and weeks 2, 4, and 6, under orthosis-assisted and non-assisted conditions, with devices iteratively adjusted over six weeks. Satisfaction was evaluated at follow-up completion using the Quebec User Evaluation of Satisfaction with Assistive Technology (QUEST 2.0-TR), and data were analyzed using a linear mixed model. Of the 13 individuals screened, 10 completed the protocol (one discontinued owing to discomfort and two withdrew voluntarily); no serious adverse events occurred. The CVA was significantly higher, and the TKA significantly lower, under the orthosis condition at every time point (all p < 0.001), reflecting an acute, device-on effect rather than a demonstrated lasting, unassisted postural correction. Mean device satisfaction was 4.88 ± 0.12/5 and mean service satisfaction was 5.00 ± 0.00/5 among the 10 completers; these scores exclude the participant who discontinued owing to discomfort and two who withdrew for unrelated reasons, which likely inflates the apparent satisfaction level. The custom-manufactured orthosis was feasible to design and fabricate, tolerable after iterative adjustment, and associated with improved craniocervical and thoracic alignment when worn; these findings reflect an acute, device-on effect and do not yet establish lasting postural correction without the orthosis. Given the single-arm, non-blinded pilot design, findings are preliminary and hypothesis-generating, supporting further confirmatory trials.
Background: Predictive performance does not establish whether model-identified risks correspond to feasible, equitable pathways. We assessed the actionability and equity of a severe tooth loss prediction model using constrained algorithmic recourse and same-source temporal validation. Methods: An Explainable Boosting Machine was trained using 2022 Behavioral Risk Factor Surveillance System data from 433,772 adults. High-risk adults underwent recourse auditing incorporating immutability locks, behavioral directionality, physiological safety floors, and at most three feature changes. Recourse represented hypothetical movement within model space, not treatment advice, causal risk reduction, or reversal of tooth loss. Primary reachability was the unweighted analytic-cohort proportion for which the frozen engine identified a feasible pathway; a BRFSS survey-weighted domain sensitivity analysis used final weights, strata, and primary sampling units. The audit separated reachability from conditional burden among reachable adults. Equity was evaluated across Social Indicators of Disparity Index (SIDI) and income strata using Oaxaca–Blinder decomposition. The frozen specification was evaluated in a 2022 holdout and the 2024 BRFSS cohort (N = 448,213) without retraining. Results: Survey-weighted areas under the receiver operating characteristic curve were 0.858, 0.855, and 0.858 in the 2022 full, 2022 holdout, and 2024 cohorts. Primary unweighted reachability was 10.61%, 10.54%, and 9.82%; corresponding survey-weighted estimates were 12.69% (95% design-aware CI, 12.33–13.07%), 12.31% (11.54–13.12%), and 11.30% (10.95–11.65%), respectively. Thus, reachability remained limited under both estimands. The high-SIDI group showed poorer calibration. Under the prespecified primary SIDI-neutral additive-cost specification, residual conditional cost differences across SIDI strata were small; alternative burden metrics were direction-dependent. The income residual attenuated from the 2022 holdout to 2024, although its confidence intervals were fixed-pipeline row-bootstrap intervals rather than fully design-based intervals. Conclusions: Stable predictive performance masked limited model-space actionability. Integrating explicitly labeled reachability estimands, conditional burden, equity, and temporal transport can strengthen pre-deployment evaluation. Because severe tooth loss is irreversible, recourse pathways should be interpreted as a stress test of model-implied modifiable factors rather than evidence of reversibility.
Paratuberculosis is a chronic infectious disease associated with substantial economic losses in livestock production. Histopathological examination remains an important diagnostic approach but can be time-consuming and dependent on specialist expertise, motivating the development of automated and interpretable image-classification methods. This study evaluated an explainable framework integrating a pretrained Swin-Tiny Transformer, handcrafted Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP) texture descriptors, and XGBoost classification for paratuberculosis histopathology image analysis. Following duplicate screening, 349 unique images comprising 199 MAP-positive and 150 MAP-negative samples were evaluated using stratified image-level five-fold cross-validation. Four model configurations were compared to assess the independent and incremental contributions of the learned and handcrafted feature representations. The standalone Swin-Tiny model achieved the highest mean ROC–AUC of 0.979±0.015, while the Swin-embedding XGBoost and hybrid Swin + GLCM/LBP + XGBoost models achieved mean ROC–AUC values of 0.977±0.016 and 0.977±0.017, respectively. The GLCM/LBP-only model achieved a mean ROC–AUC of 0.934±0.041, indicating that the handcrafted texture descriptors contained independently discriminative information but provided limited incremental value when combined with the Swin embeddings. Grad-CAM and XGBoost feature-importance analyses provided image-level and feature-level insights into model predictions. These findings demonstrate the effectiveness of Swin-Tiny representations for paratuberculosis histopathology image classification while highlighting the need for external validation using larger, independently sourced datasets.
Biological strategies for converting carbon dioxide (CO2) into valuable compounds are attractive approaches for a carbon-neutral future. Bioelectrochemical systems (BESs) represent an innovative strategy for the control of microbial metabolism, in which electrochemical techniques are adopted to stimulate reductive and oxidative processes. Acetogenesis and methanogenesis are the two main chemoautotrophic pathways of CO2 reduction usually present in anaerobic environments. Due to the syntrophic and competitive relationship between acetogens and methanogens, methanogenesis inhibition strategies should be adopted to direct CO2 reduction towards acetate and fatty acids. In this work, an acetogen-enriched inoculum was produced by the bioaugmentation of Acetobacterium woodii in the heat-shocked and acid-treated inoculum. Then, by using H-cell reactors, without the use of any chemical inhibitor, this inoculum was tested in semi-continuous mode by imposing a dilution rate previously identified from growth kinetic assessment. Bioelectrochemical tests, conducted at −0.9 V and −0.7 V vs. SHE, showed the overcoming of acetogenesis on methanogenesis. At −0.7 V vs. SHE, acetate was produced at 0.0385 ± 0.009 mmol d−1 and 0.0343 ± 0.010 mmol d−1 in the absence and presence of bioaugmentation, respectively, with acetate cathodic coulombic efficiency (CCEs) of 68% and 64%. At −0.9 V vs. SHE, bioaugmentation markedly reduced methanogenesis, decreasing the methane production rate from 0.105 ± 0.012 to 0.007 ± 0.004 mmol d−1, while acetate production reached 0.061 ± 0.025 mmol d−1 with a CCE of 43%. Finally, the effect of bioaugmentation was demonstrated by cyclic voltammetry of the biocathode, which showed the increase in biocatalytic activity due to the presence of Acetobacterium woodii.
Peripheral neuropathy (PN) is a debilitating condition characterized by chronic pain, numbness, and motor dysfunction, with limited treatment options. Ischemic stroke can cause central neuropathy, which may also induce PN. Human mesenchymal stem cells (hMSCs) have shown promise in therapeutic applications, but limitations in cell viability, immune response, and efficacy persist. Extracellular vesicles (EVs), which facilitate cell-free intercellular communication, offer a promising alternative for nerve regeneration. Electrical stimulation (ES) has emerged as a method to enhance EV secretion, and this study investigates its potential for promoting EV production from human adipose tissue-derived mesenchymal stem cells (hASCs) and human Schwann cells (hSCs). In this study, hASCs, hSCs, and lipopolysaccharide (LPS)-induced inflamed hSCs were subjected to one hour of low-frequency direct current (DC) electrical stimulation (100 mV/mL) for 7 days. EVs were isolated using differential ultracentrifugation and characterized through nanoparticle tracking analysis (NTA). Gene expression was analyzed via qRT-PCR to evaluate markers associated with EV biogenesis as well as pro- and anti-inflammatory cytokines. Our results demonstrate that ES significantly increases EV secretion from both hASCs and hSCs, with a notable upregulation of genes involved in both the endosomal sorting complex required for transport (ESCRT)-dependent and ESCRT-independent pathways of EV biogenesis. Additionally, ES modulates inflammation-related markers, promoting anti-inflammatory gene expression and reducing pro-inflammatory gene levels. Notably, LPS-induced hSCs exhibited a phenotype shift from myelinating to non-myelinating cells, producing EVs capable of modulating the inflammatory microenvironment. However, prolonged exposure to ES led to a decrease in EV secretion and changes in EV size distribution, suggesting potential cellular adaptation or membrane stress. This study highlights the potential of ES as a scalable, cell-free strategy to enhance EV production, offering new insights into its therapeutic applications for peripheral neuropathy and nerve regeneration.
Urinary incontinence (UI), the involuntary leakage of urine, is a common condition that imposes significant physical, emotional, and financial burdens. Injury to the external urethral sphincter (EUS) contributes to myogenic UI by damaging the skeletal muscle responsible for urinary control. This study evaluated the engraftment and therapeutic potential of human muscle-derived cells (hMDCs) using acute and chronic skeletal muscle injury models. Immunocompromised female rats received cardiotoxin (CTX) or volumetric muscle loss (VML) injuries to the tibialis anterior muscle. Animals were treated with 100,000 or 500,000 passage 3 or passage 6 hMDCs injected one day after CTX injury or six weeks after VML injury. VivoTrack imaging and human leukocyte antigen staining confirmed successful engraftment of hMDCs at the injection site. A total of 500,000 passage 3 hMDCs reduced muscle force, whereas later-passage 6 cells had no effect on force production. All treatment groups exhibited small, newly regenerated muscle fibers. In chronic VML injuries, hMDC engraftment using 500,000 passage 6 cells enhanced regenerative characteristics despite limited fusion with host muscle fibers. These findings suggest that hMDCs improved regenerative characteristics in a fibrotic VML model but did not restore contractile force. This work supports the therapeutic potential of hMDCs and provides a foundation for future clinical strategies to treat urinary incontinence.
Microalgae are promising platforms for biomass production, carbon capture, biofuels, and high-value bioproducts. However, despite significant advances in cultivation technologies, reactor engineering, and metabolic engineering, industrial implementation remains limited. This gap suggests that the main challenge of microalgae biotechnology lies not in the availability of productive strains or cultivation systems, but in managing the environmental and physiological heterogeneity that emerges during scale-up. This structured narrative review selected literature using predefined descriptors and relevance-based inclusion criteria and organized the evidence into five thematic domains encompassing cultivation-scale constraints, cellular physiology, bioengineering, precision technologies, and industrial translation. This review examines macrospatial bottlenecks related to light distribution, gas transfer, hydrodynamics, and reactor operation, alongside microspatial constraints involving cell cycle regulation, carbon allocation, metabolic adaptation, and stress responses. Recent advances in adaptive cultivation, real-time monitoring, artificial intelligence, digital twins, computational modeling, and bioengineering are discussed as tools to transform biological and environmental variability into actionable information. Based on concepts established in precision agriculture, this review proposes precision microalgae as a conceptual framework that integrates reactor engineering and cell physiology with three operational pillars: real-time monitoring, predictive modeling, and adaptive control. Its specific contribution is to connect currently fragmented technological and biological advances within a common framework for managing multiscale heterogeneity during cultivation and scale-up. Overall, the available evidence supports the operational logic of this framework, although its generalized effectiveness under industrial conditions remains to be demonstrated.
Alzheimer’s disease and related dementias are projected to affect more than 150 million people worldwide by 2050. Early staging with validated instruments such as the Clinical Dementia Rating (CDR) scale is essential for timely intervention, yet access to clinician-administered CDR assessment remains constrained by workforce, time, and geographic barriers. This study complements a previously published machine learning pipeline for Alzheimer’s disease prediction by addressing the downstream task of dementia staging. Because the global CDR score is already derived from the six sub-domain ratings through an established rule-based procedure, the contribution reported here lies not in discovering that mapping but in encoding it in a transparent, deployable form: an explainable decision tree classifier embedded in DiAbot, a large-language-model-fronted conversational system that supports self-administered CDR-style assessment. We extracted 13,453 CDR records from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), removed administrative variables, invalid entries, and missing rows (final n = 13,290), and trained decision tree classifiers under two impurity criteria, Information Gain and Gini Index, using a 70/30 stratified record-level hold-out and ten-fold stratified record-level cross-validation. This classifier-level evaluation uses the six domain scores as recorded during ADNI’s clinician-administered assessment, not scores elicited by the DiAbot chatbot; the trained classifier was separately embedded in a web application in which a prompt-engineered large language model conducts a CDR-style interview and normalizes responses to ordinal domain scores, but the end-to-end accuracy of that full conversational pipeline (chatbot elicitation through to final CDGLOBAL) has not yet been measured, and is not what the headline accuracy figures below report. The Information Gain Decision Tree reproduced the established mapping from the six CDR sub-domain scores to the CDGLOBAL with 99.86% accuracy under the record-level hold-out protocol (matching macro-averaged precision, recall, and F1-score), with a ten-fold record-level cross-validated mean of 99.81% (SD 0.07); this result represents fidelity to the established CDR scoring rule rather than independent dementia-diagnosis accuracy. Gini-based trees performed almost identically (99.79% hold-out, 99.74% cross-validated). Memory dominated feature importance, consistent with its role as the primary domain in the official CDR scoring algorithm. Residual misclassifications were confined to adjacent CDR stages. Because the CDGLOBAL is deterministically derived from the six sub-domain scores, these figures should be read throughout as evidence of high-fidelity reconstruction of the established CDR scoring relationship, not as general dementia-diagnosis accuracy comparable to imaging- or biomarker-based classifiers; further, participant-independent generalization remains unverified under the record-level protocol evaluated here. An interpretable classifier embedded in a conversational front-end can nonetheless make standardized CDR-style staging more widely accessible while preserving clinical inspectability; the resulting system is positioned as a screening-stage adjunct to, and not a replacement for, clinician-administered CDR assessment.
Adult acquired flatfoot deformity (AAFD) results from the collapse of the medial longitudinal arch (MLA), often managed conservatively with foot orthoses. While 3D printing enables rapid and cost-effective production of insoles, the biomechanical impact of different printing materials remains unclear. This study combined material testing, finite element (FE) simulation, and compression testing to evaluate four materials commonly used for 3D-printed insoles: PLA (polylactic acid), resin, EVA (ethylene vinyl acetate), and TPU (thermoplastic polyurethane). Tensile tests provided material properties for FE modeling of balanced standing using a validated healthy foot model. Outcome measures included MLA deformation (navicular drop), plantar stress distribution, and von Mises stress within the insole. A commercial semi-custom insole was tested for comparison. PLA and resin exhibited high stiffness (2228 and 880 MPa) and reduced navicular drop by 55–60%, compared with <30% reductions by EVA and TPU. Plantar stress distribution shifted from forefoot and hindfoot toward the midfoot, with PLA increasing midfoot load share to 38%. Compression testing confirmed that PLA and resin exhibited greater arch support ability than the semi-custom insole within a 10 mm displacement threshold. Material stiffness strongly influenced the biomechanical performance of 3D-printed insoles. Within this static simulation baseline, PLA and resin provided superior arch stabilization and stress redistribution compared with EVA, TPU, and a semi-custom reference. These findings suggest that material choice is critical to optimizing 3D-printed orthotic support, providing a valuable biomechanical baseline to guide future orthotic designs for flatfoot.
Multimodal learning has gained attention in recent years due to its ability to effectively utilize data features from various modalities. Diagnosing the vulnerability of atherosclerotic plaques directly from carotid 3D MRI images is challenging for both radiologists and conventional 3D vision networks. In clinical practice, radiologists assess patients using a multimodal approach that incorporates various imaging modalities and domain-specific expertise, paving the way for the creation of multimodal diagnostic networks. In this study, we proposed an effective framework to leverage radiologists’ domain knowledge to improve the automated diagnosis of carotid plaque vulnerability through variational inference and multimodal knowledge distillation (VMD). This framework excels in harnessing cross-modality prior knowledge from limited image annotations and radiology reports within training data, thereby enhancing the diagnostic network’s accuracy for unannotated 3D MRI images. We validated the proposed VMD framework on our in-house dataset, demonstrating its effectiveness.
Background: Uro-Dynamic MRI is a non-invasive protocol used to evaluate/analyze bladder contraction. Regionalized anatomical measurements and multi-variate linear regression models (MVLRM) were used to establish a methodology to evaluate bladder contraction. Methods: In this exploratory study, five healthy subjects and six LUTS patients underwent Uro-Dynamic MRI during voiding. Bladders were segmented at each time point during the void. Volume (V) and surface area (SA) were recorded, resulting in urinary flow rate (Q) and sphericity index (SI) curves. Regionalized anatomical measurements were tracked and linearly correlated with Q, while symptom scores from the International Prostate Symptom Score Survey (IPSS) were correlated using MVLRM. Results: Healthy subjects exhibited high linear correlation of contraction with Q; the opposite was observed in LUTS patients. The occurrence of SImax coincided within two time points of Qmax for 80% of healthy subjects and 50% of LUTS patients. MVLRM identified SI, transverse and anteroposterior lengths as correlates of LUTS diagnosis and urinary frequency. Conclusions: This methodology revealed distinct, biphasic patterns of bladder contraction during voiding and showed associations between symptomatic behavior and anatomical measurements. This exploratory study demonstrated a comprehensive methodology to non-invasively quantify patient-specific bladder contraction patterns during voiding that could provide physicians with relevant information on symptom-specific treatment.
Skeletal-anchored mesial sliders are increasingly used for molar mesialization, yet their mechanical behavior has not been compared under standardized conditions. This in vitro pilot study evaluated four slider designs (BENEfit® Beneslider [BB], TADMAN Beneslider [TB], IZE Slider [IO; OrthoLIZE GmbH], Slider on Minipin [SO; OrthoLIZE GmbH]), each combined with elastic chains (-C) or NiTi springs (-S), to characterize force loss, mesial movement, and associated mechanical side effects using an experimental biomechanical measurement system. Using repeated measurements on one specimen with 1 N of applied force, three-dimensional tooth movements were recorded across 200 simulation steps per run. The sliders showed design-dependent mechanical patterns: IO sliders produced the smallest sagittal movements (IO-C: -0.84 mm [0.05], IO-S: -0.79 mm [0.02]) and the highest force loss (IO-C: 88.1% [2.1], IO-S: 89.8% [0.3]), whereas BB-S and TB-S generated larger mesialization distances (BB-S: -3.55 mm [0.53], TB-S: -3.46 mm [0.08]) with lower force loss (BB-S: 42.8% [9.2], TB-S: 21.1% [2.9]). Translational deviations remained small, while rotational effects were more pronounced. These findings represent mechanical tendencies of the tested configurations under idealized conditions and do not account for biological variability, periodontal compliance, or patient-specific factors. As such, the results cannot be generalized to clinical performance but provide preliminary reference data. Expanded investigations using multiple specimens, biological modeling, or finite element analysis will be necessary to determine how these mechanical patterns translate into clinical tooth movement.
Individuals with transtibial amputation have impaired postural control due to limited ankle function. Conventional energy-storing-and-return (ESAR) prosthetic feet mainly support sagittal-plane mechanics and provide limited frontal-plane adaptability. This pilot pre-post motion analysis study examined the effects of the Intersection Foot® (IF), a torsion-adaptive ESAR prosthetic foot, on balance control in 10 male individuals with unilateral transtibial amputation. Three-dimensional motion analysis and force plate measurements were performed during five Berg Balance Scale-derived tasks. During sit-to-stand, IF reduced the mediolateral root mean square distance of center of pressure from 10.20 ± 3.59 to 8.29 ± 3.32 mm (p = 0.047) and the anteroposterior center of mass (CoM) range from 345.51 ± 31.72 to 326.45 ± 47.40 mm (p = 0.008). During tandem stance, IF increased mediolateral CoM range from 23.94 ± 18.77 to 55.99 ± 30.90 mm (p = 0.021), vertical CoM range from 4.14 ± 4.18 to 9.45 ± 7.72 mm (p = 0.038), and sound-side frontal-plane ankle ROM from 1.44 ± 1.20° to 2.48 ± 1.30° (p = 0.022). Overall, replacing the habitual feet with IF produced task-dependent biomechanical adaptations rather than uniform improvements in postural stability, suggesting reduced dynamic sway during sit-to-stand but increased whole-body and sound-limb compensatory motion during tandem stance.
Accurate morphological analysis of blood smears is vital for hematological diagnosis, yet manual examination is labor-intensive and subjective. While deep learning offers automation, its black-box nature and computational demands often hinder clinical trust and deployment. We propose XHIC-Net, an Explainable Hybrid Involution-Convolution Network designed for efficient and transparent cell classification. By integrating spatially adaptive involution operations with convolutional layers within a residual framework, XHIC-Net captures both contextual and fine-grained features efficiently. To enhance interpretability, a Grad-CAM-based explainable AI (XAI) module visualizes the cellular regions driving model predictions. The proposed framework was evaluated on a dataset comprising 12,879 microscopic blood smear images belonging to 12 morphological cell categories. Experimental results demonstrate that XHIC-Net achieves an overall accuracy of 98.88%, precision of 98.89%, recall of 98.87%, F1-score of 0.9887, and Cohen's Kappa score of 0.9887. It outperformed established models, including DL models such as EfficientNetV2S, MobileNet family, DenseNet family, and VGG16, while using fewer parameters and requiring shorter training times. Furthermore, the XAI maps consistently highlighted biologically relevant structures, validating the model's decision-making process. XHIC-Net is a strong, effective, and clear research model for automated hematology. With future clinical validation, it has the potential to be modified for point-of-care diagnostics in healthcare settings with limited resources.
BACKGROUND:The ASA Physical Status (ASA-PS) classification and the Charlson Comorbidity Index (CCI) are common pre-operative scoring tools. Language models could automate structured pre-operative scoring, but direct comparisons require paired inference because all models are evaluated on the same patients. METHODS:In this retrospective single-center concordance analysis, 101 consecutive adult orthopedic patients were independently rated by two clinicians; the rounded mean for ASA-PS and arithmetic mean for CCI formed a clinician-derived composite reference. The cohort contained no ASA-PS IV-V patients. Six model configurations received identical prompts. Agreement was assessed using quadratic weighted kappa, ICC(2,1), exact and adjacent agreement, MAD, RMSE, and Bland-Altman limits. Post hoc between-model comparisons used 10,000 patient-level paired bootstrap replicates with Benjamini-Hochberg correction. RESULTS:Inter-clinician weighted kappa was 0.713 for ASA-PS and 0.914 for CCI. GPT-5.2 reached kappa 0.884 for ASA-PS and 0.970 for CCI. In paired analyses, GPT-5.2 had significantly higher quadratic weighted kappa than every other tested model for both outcomes and significantly higher ICC for CCI after multiplicity correction. Phi4 and deepseek-r1-70B were not significantly different from inter-clinician agreement for CCI kappa or ICC; equivalence was not tested. CONCLUSIONS:Among the six evaluated model configurations, GPT-5.2 achieved significantly higher agreement with the clinician-derived composite reference than the other tested models for both ASA-PS and CCI in post hoc paired analyses with multiplicity correction. Locally deployable phi4 and deepseek-r1-70B showed CCI agreement estimates that were not statistically distinguishable from inter-clinician agreement, although equivalence was not tested. These findings are limited to the evaluated models and study cohort.
Artificial intelligence and bioengineering have developed along different historical trajectories, yet their areas of interest have progressively converged [...]
Early identification of COVID-19 patients requiring intensive care is critical for improving treatment prioritization, supporting clinical decision-making, and managing limited hospital resources. While structured electronic health record (EHR) data provide important physiological information, chest computed tomography (CT) imaging contains additional indicators related to disease severity. This study presents a multimodal clinical decision support framework for a multimodal risk prediction framework for severe COVID-19 outcomes using structured emergency clinical features and patient-level CT imaging from the COVID Data for Shared Learning (CDSL) dataset. After multimodal cohort construction, 784 patients with both structured clinical records and CT imaging were included in the analysis. Three predictive settings were evaluated: EHR-only prediction using Gradient Boosting, CT-only prediction using ResNet50-based feature extraction with Logistic Regression, and multimodal prediction using weighted late fusion. The experimental results indicate that the CT-based model surpassed the clinical baseline, yielding an F1-score of 0.42 and an ROC-AUC of 0.772, whereas the EHR-only model achieved scores of 0.30 and 0.715, respectively. Overall, the multimodal fusion framework achieved the strongest results among the approaches tested, reaching an F1-score of 0.47 and an ROC-AUC of 0.782. Taken together, these findings indicate that, although CT imaging alone carries meaningful predictive power for evaluating ICU risk, combining it with clinical data leads to predictions that are more reliable and robust. The proposed framework offers a practical and interpretable foundation for multimodal clinical decision support and demonstrates the potential of combining structured clinical data with medical imaging for intelligent critical care applications.
