Search Results (10,746)

Background and Objective: When hypertensive cerebral hemorrhage causes secondary intraventricular hemorrhage, it usually significantly increases the complexity of the patient’s condition and the risk of poor prognosis. Deep learning methods can automatically and quickly segment intraparenchymal hemorrhage and intraventricular hemorrhage, and quantitatively analyze hematoma-related properties to provide auxiliary information for subsequent diagnosis and treatment. Methods: We retrospectively enrolled 351 patients with hypertensive cerebral hemorrhage, 219 of whom had secondary intraventricular hemorrhage. All patients underwent computed tomography within 1 week after diagnosis. Based on 3D U-Net, we developed a deep learning network with a multi-scale deformable convolution module and a softened anatomical consistency loss. The multi-scale deformable convolution module can enhance the learning ability of multi-deformation features and increase the receptive field of the network. The anatomical consistency loss, built upon softened labels, can alleviate the impact of label noise. Results: We evaluated our model at pixel, volume, and morphology levels. It achieved Dice of 0.8990 ± 0.1169 for intraparenchymal hemorrhage and 0.7124 ± 0.1227 for intraventricular hemorrhage, both higher than that of the comparison model. Compared with the Coniglobus method, our model has a narrower consistency limit and more concentrated predicted values. Additionally, in segmenting irregular and different-sized hematomas, it generates the smallest centroid, volume, position, and morphology deviations. Conclusions: The proposed model can automatically and accurately segment two types of hematomas and quantify multiple attributes, is robust in multi-deformation and label noise scenarios, and has the potential to assist in clinical diagnosis and treatment; its actual impact on clinical decision-making and patient outcomes requires prospective validation. Full article

With immune checkpoint inhibitors becoming the mainstay of systemic therapy in both metastatic and early-stage settings across many cancer types, the management of immune-related adverse events has emerged as a central priority of modern oncological supportive care. These toxicities can substantially impair the quality of life of cancer patients, including those who achieve long-term survival. Consequently, there is a pressing need to develop reliable predictive biomarkers to better tailor immune checkpoint inhibitor treatment and optimize patient selection. This review summarizes several of the most promising predictive biomarkers currently under investigation, including genetic factors; peripheral blood parameters and their ratios; autoantibodies; cytokines and chemokines; cytomegalovirus serostatus; gut microbiome characteristics; body composition metrics; molecular imaging features; and tumour- and patient-related factors such as cancer type, gender, and physical activity. Because single biomarkers have limited predictive value, multi-omics prediction models and composite immune-cell scores are increasingly demonstrating greater potential. However, none of these candidate biomarkers have yet undergone sufficient validation to support their incorporation into routine clinical practice. Full article

Acute pancreatitis is a common and heterogeneous inflammatory disorder whose clinical course ranges from a self-limited illness to persistent organ failure, infected pancreatic necrosis, and prolonged critical illness. Contemporary management has moved away from protocolised aggressive fluid loading, prolonged fasting, prophylactic antibiotics, and early open necrosectomy. Instead, current care emphasises repeated physiological assessment, moderate goal-directed resuscitation, early enteral or oral nutrition, organ-specific support, antimicrobial stewardship, and delayed minimally invasive intervention within a multidisciplinary step-up strategy. This narrative review examines acute pancreatitis from an intensive care perspective. Particular attention is given to early risk stratification, intensive care unit triage, haemodynamic and respiratory support, acute kidney injury, intra-abdominal hypertension, nutrition, biliary source control, diagnosis and treatment of infected necrosis, and the timing and selection of endoscopic, radiological, and surgical interventions. The implications of obesity, pregnancy, advanced age, and multimorbidity are also discussed. Recent randomised trials have clarified several clinically important questions: aggressive hydration increases fluid overload without improving outcomes; routine urgent endoscopic retrograde cholangiopancreatography is not beneficial in predicted severe biliary pancreatitis without cholangitis; postponed drainage may avoid invasive intervention in a substantial proportion of patients with infected necrosis; and endoscopic or minimally invasive approaches reduce treatment burden compared with primary open surgery. Persistent organ failure remains the principal determinant of mortality, while infected necrosis further increases risk and complexity. Future progress will depend on dynamic prediction models, biomarker-guided antimicrobial decisions, personalised haemodynamic strategies, phenotype-directed immunomodulation, and regionalised multidisciplinary care. Full article

Background: Portal vein tumor thrombus (PVTT) marks advanced hepatocellular carcinoma (HCC) and carries a dismal prognosis. Survival varies widely even within this stage, yet simple tools for individualized risk stratification remain scarce. Methods: We retrospectively enrolled 134 HCC patients with PVTT and randomly divided them into a training set (n = 94) and a validation set (n = 40). Clinical predictors were selected by variance inflation factor screening and backward elimination Cox regression. A radiomics score (Rad-score) was constructed from portal-venous phase computed tomography (CT) images using Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression with 10-fold cross-validation. Three Cox models were built: a clinical model, an imaging model based solely on the Rad-score, and a combined model integrating both. Discrimination was assessed by C-index and time-dependent area under the curve (AUC). Calibration was examined with bootstrap-based calibration curves. Decision curve analysis evaluated net benefit. A nomogram was developed from the combined model. Results: Four clinical variables (alpha-fetoprotein (AFP), body mass index (BMI), high-density lipoprotein cholesterol (HDL-C), and alkaline phosphatase (ALP)) and two CT texture features (GLRLM_SRHGE and GLZLM_SZHGE) were retained as independent predictors. The combined model gave the highest C-index in both the training set (0.843) and the internal validation set (0.815). Its 1-year AUC reached 0.953 and 0.947 in the two sets. Calibration slopes ranged from 1.044 to 1.291 across time points, indicating a tendency toward mild overdispersion; nevertheless, decision curve analysis confirmed net benefit across clinically relevant thresholds. The combined model offered greater net benefit than either single-domain model across a 0–50% threshold range. A nomogram incorporating all five predictors was generated for individualized 12- and 24-month survival prediction. Conclusions: A combined model integrating routine laboratory variables and a CT-based radiomics score improved survival prediction over clinical or imaging models alone. The corresponding nomogram uses inputs from a basic blood panel and a single portal-venous phase CT, suggesting its potential as a low-cost prognostic stratification tool for HCC patients with PVTT, although external validation in prospective multicenter cohorts is required before clinical implementation. Full article

Background: Artificial intelligence (AI) has expanded rapidly across orthopaedic practice, yet routine clinical adoption remains limited despite strong technical performance. This narrative review examines why a persistent gap separates technical maturity from clinical maturity across the orthopaedic patient care pathway. Methods: We performed a structured qualitative evidence synthesis of contemporary high-level evidence (systematic reviews, diagnostic test accuracy meta-analyses, and structured narrative reviews) retrieved from PubMed/MEDLINE, Scopus, and Web of Science, supplemented by backward screening of reference lists, covering January 2022 to June 2026. Twenty-one evidence syntheses were analysed thematically across six predefined analytical domains and organized according to the orthopaedic patient pathway. Reporting followed the SANRA (Scale for the Assessment of Narrative Review Articles) criteria. Results: Musculoskeletal imaging and fracture detection represented the most mature domains, with several applications reaching early clinical adoption. Applications in arthroplasty planning, shoulder surgery, perioperative prediction, multimodal AI, and clinical decision support remained at developing or emerging stages. Recurrent barriers included limited external validation, dataset heterogeneity, poor workflow interoperability, limited explainability, regulatory and ethical uncertainty, and scarce patient-centred outcome evidence. Conclusions: The principal challenge facing orthopaedic AI is no longer algorithm development but clinical translation. We propose the ORION Clinical Readiness Framework, an evidence-informed five-domain model describing the transition from Technical Performance through Clinical Validation, Workflow Integration, and Patient Benefit to Routine Clinical Adoption, to guide implementation and future research. Full article

Background/Objectives: Postoperative atrial fibrillation (POAF) is a common complication after cardiac surgery with cardiopulmonary bypass (CPB), increasing morbidity and prolonging hospitalization. This study aimed to develop and validate an exploratory prediction model that integrates perioperative inflammatory biomarkers with clinical and surgical variables to identify patients at risk of early POAF. Methods: A prospective exploratory cohort of 89 patients undergoing coronary artery bypass grafting (CABG; n = 36), valve surgery (n = 40), or CABG–valve surgery (n = 13) was evaluated. Clinical, surgical, and proinflammatory serum biomarkers (IL-6, IL-8, IL-10, and CRP) were recorded preoperatively (T1) and at 24 h (T2) and 48 h (T3) postoperatively. Multiple-comparison adjustments were made using the Benjamini–Hochberg false discovery rate. Predictor selection was based on bootstrap-derived stability using LASSO-penalized logistic regression, and the final model was estimated using Firth’s bias-reduced logistic regression. Results: POAF incidence was 8.3% in CABG, in contrast to 22.5% and 30.8% in valve and CABG-valve surgeries, respectively. After multiple-comparison corrections, only IL-6 at T2 postoperatively was significantly higher in patients who subsequently developed POAF. Bootstrap-based stability selection retained T2 postoperative IL-10 and magnesium concentrations in the final model, which achieved an apparent AUC of 0.776 and a bootstrap optimism-corrected AUC of 0.728, with acceptable calibration (Brier score = 0.103), negligible multicollinearity (VIF = 1.04), and a negative predictive value of 95.5% at the optimal Youden threshold. Conclusions: Our findings support an exploratory prediction model with moderate discrimination for POAF after cardiac surgery with CPB, providing a methodological foundation for future multicenter validation studies. Full article

Background/Objectives: Carotid revascularization reduces stroke risk in patients with carotid stenosis. Postoperative length of stay (LOS) reflects resource utilization and recovery, yet its determinants remain poorly described in Central Asia. This study examined whether baseline patient characteristics predict prolonged postoperative LOS in a Kazakhstani referral cohort. Methods: We analyzed a single-center retrospective cohort (clinical registry) of 320 consecutive patients who underwent carotid revascularization between January 2018 and December 2025. Carotid endarterectomy (CEA) was performed in 88 patients (27.5%) and carotid artery stenting (CAS) in 228 (71.2%). Prolonged postoperative LOS was defined a priori as >8 days (75th percentile). Multivariable logistic regression was performed, with negative binomial regression; a model additionally including procedure type, and a model including in-hospital complications were used as sensitivity analyses. Results: Patients were predominantly male (74.7%) with a mean age of 71.1 ± 7.5 years and a high burden of comorbidities. Median postoperative LOS was 6 days (IQR 4–8), and 72 patients (22.5%) had prolonged LOS. Postoperative LOS was similar after endarterectomy and stenting (median 6 vs 5 days). In-hospital complications occurred in 11 patients (3.4%), with two deaths (0.6%). No baseline characteristic independently predicted prolonged LOS (all p > 0.05; AUC = 0.629; Hosmer–Lemeshow p = 0.45), and procedure type was not associated with prolonged LOS (adjusted OR 1.14, 95% CI 0.61–2.14). In contrast, in-hospital complications were strongly associated with prolonged LOS (adjusted OR 8.50, 95% CI 2.04–35.40; p = 0.003), although this estimate was imprecise owing to the small number of events. Median postoperative LOS increased from 6 days in patients without complications to 12.5 days in those with complications (p < 0.001). Conclusions: In this elderly, comorbidity-heavy cohort, prolonged postoperative LOS was associated with perioperative complications rather than baseline patient characteristics. Because complications lie on the causal pathway between baseline risk and hospital stay, this finding is best interpreted as hypothesis-generating: it suggests that efforts to shorten stay may be better directed toward complication prevention than preoperative risk stratification, a hypothesis that warrants prospective evaluation. Full article

Neurodegenerative diseases such as Alzheimer’s disease (AD) and Parkinson’s disease (PD) develop over decades, yet their earliest pathogenic drivers remain poorly understood. Epidemiological and experimental animal studies suggest that disturbances in oral sensorimotor regulation, particularly within trigeminal proprioceptive pathways, may contribute to neural dysfunction long before clinical symptoms emerge. The mesencephalic trigeminal nucleus (MesV), the only primary sensory neuron population located entirely within the central nervous system (CNS), links oral proprioception with brainstem and forebrain networks. Chronic occlusal mismatch, impaired mastication, sleep bruxism, and sleep-disordered breathing may generate persistent sensorimotor prediction errors that destabilize MesV-centered circuits and subsequently recruit the locus coeruleus (LC), the brain’s principal noradrenergic stress nucleus. This review proposes an oral–brain axis model in which chronic MesV-related prediction error signaling engages LC-dependent stress systems, leading to neuroimmune activation, locus coeruleus–asparagine endopeptidase (LC-AEP) pathway engagement, and downstream proteinopathic processes. Sustained LC activity may facilitate microglial priming, reactive astrocytosis, and neuroinflammatory signaling, creating conditions that favor LC-AEP pathway activation and downstream tau pathology. Epidemiological studies associate tooth loss, reduced occlusal support, and impaired mastication with increased dementia risk, while experimental models of prodromal PD demonstrate early trigeminal sensory-processing abnormalities preceding motor symptoms. Together, these findings support the hypothesis that chronic disturbances in oral sensorimotor homeostasis may increase neurodegenerative vulnerability. This framework identifies potential biomarkers and preventive targets, suggesting that modulation of oral function and neuroimmune pathways may help reduce neurodegenerative risk before irreversible neuronal loss occurs. Full article

Background: Central arterial stiffness assessed by carotid–femoral pulse wave velocity (PWVc-f) strongly predicts cardiovascular risk. We examined associations of objectively measured physical activity, sedentary behavior, and functional fitness with PWVc-f in community-dwelling older adults. Methods: The study included 170 older adults (124 women; 64–91 years). Moderate-to-vigorous physical activity (MVPA) and sedentary behavior were assessed by accelerometry, functional fitness by the Fullerton Functional Fitness Test (composite score), and PWVc-f by applanation tonometry. Associations were examined using correlation, hierarchical linear regression, and logistic regression analyses. Results: Elevated arterial stiffness (PWVc-f > 10 m/s) was present in 62.3% of participants. PWVc-f was associated with all functional fitness measures, including the composite score (ρ = −0.338, p < 0.001), but not with MVPA or sedentary behavior. High functional fitness was associated with lower odds of elevated PWVc-f in unadjusted analyses (OR = 0.39, 95% CI 0.20–0.78; p = 0.007), but this was no longer significant after adjustment for age, sex, and anthropometric/clinical characteristics. Neither MVPA nor sedentary behavior predicted PWVc-f. Age remained the only independent correlate across all models. Conclusions: Functional fitness showed stronger unadjusted associations with PWVc-f than MVPA or sedentary behavior. However, after multivariable adjustment including sex and clinical covariates, age emerged as the only consistent independent correlate of PWVc-f. These findings highlight the importance of considering age when interpreting the relationship between functional fitness and arterial stiffness in older adults. Full article

Triple-negative breast cancer (TNBC) remains a major therapeutic challenge due to pronounced molecular heterogeneity, transcriptional plasticity, and frequent treatment resistance. MicroRNA (miRNA)-based strategies have emerged as potential approaches for modulating dysregulated gene expression networks in TNBC; however, the contribution of understudied miRNA families to TNBC-associated regulatory programs remains incompletely understood. This study aimed to investigate the tumor-suppressive role of miR-548f-3p in TNBC and to identify candidate downstream effectors, with particular focus on ANP32E. An integrative analysis combining public transcriptomic datasets, clinical expression profiling, computational target prediction, network-based prioritization, pathway analysis, and single-cell transcriptomic assessment identified miR-548f-3p as consistently downregulated in TNBC. Among candidate downstream targets, ANP32E, a chromatin-associated regulator involved in H2A.Z histone variant dynamics, was identified as a potential effector exhibiting increased expression in TNBC and enrichment within malignant epithelial cell populations. An inverse association between miR-548f-3p and ANP32E expression was observed in patient-derived samples. In breast cancer cell models, miR-548f-3p mimic restoration increased apoptosis, promoted G₀/G₁ accumulation, and reduced migration- and invasion-associated readouts, although measurable effects were also observed in non-tumorigenic MCF-10A cells. These phenotypic changes were accompanied by reduced ANP32E expression at the protein level, indicating that ANP32E expression is responsive to miR-548f-3p restoration. This study supports miR-548f-3p as a candidate tumor-suppressive miRNA in TNBC. The reduction in ANP32E protein expression following miR-548f-3p restoration, together with computational, single-cell, and clinical expression evidence, supports ANP32E as an expression-responsive candidate downstream effector of miR-548f-3p. Further reporter-based and rescue experiments are required to confirm direct 3′UTR-mediated targeting and to define the mechanistic contribution of ANP32E within the broader miR-548f-3p regulatory network. Full article

Blood transfusion near the end of life for patients with cancer may relieve symptoms attributed to anaemia or control bleeding, but its palliative value depends on whether benefit occurs within the patient’s expected survival and outweighs the clinical and practical burden of treatment. This structured narrative review examines evidence concerning red-cell and platelet transfusion, transfusion-sparing patient blood management, hospice access and alternative delivery models in adults receiving palliative or end-of-life care. Structured searches of MEDLINE, Embase, Scopus and the Cochrane Library identified original studies, audits, qualitative research and relevant evidence syntheses. The evidence was predominantly observational and methodologically heterogeneous. Reported symptomatic response rates after red-cell transfusion ranged from 31% to 70%, most commonly involving short-term improvement in fatigue, dyspnoea or general well-being. Benefit often diminished within 14 days, while 23–35% of participants in historical cohorts died within two weeks, limiting the opportunity to experience benefit. Response was not reliably predicted by haemoglobin concentration. Red-cell transfusions were frequently guided by laboratory values, administered late in the disease course and not followed by systematic reassessment. Evidence concerning platelet transfusion remains largely descriptive and does not establish a prophylactic threshold for end-of-life care. Intravenous iron and haemostatic radiotherapy may reduce transfusion requirements in selected patients, although their applicability depends on clinical stability, the source of bleeding and sufficient time to benefit. In haematological malignancies, transfusion dependence has been associated with lower hospice enrolment and shorter hospice stays, whereas home- and hospice-based transfusion may reduce travel and waiting burdens for appropriately selected patients. Transfusion near the end of life should therefore be considered a goal-concordant, time-limited trial. A patient-valued outcome should be defined beforehand, the fewest units likely to achieve that outcome should be administered, and benefit should be reassessed using the same measure. Further transfusion should be offered only when documented symptomatic or functional benefit outweighs adverse effects and treatment burden for the individual patient. Full article

Hybrid CNN–Transformer (HCT) synthesis for automated neurodevelopmental diagnostics is an effective approach to constructing intelligent detection systems that are not merely oriented toward feature classification but primarily toward solving spatiotemporal pattern recognition problems in motor disorder assessment. In neurodevelopmental diagnostics, existing automated methods rely on fixed, single-model architectures that process spatial or temporal motion features independently, failing to adapt to the heterogeneous motor irregularities characteristic of developmental coordination disorder and degrading detection sensitivity and generalization across diverse patient populations. There is therefore a pressing need for models capable of simultaneously capturing intra-frame spatial coordination patterns and inter-frame temporal movement dependencies against interrelated diagnostic criteria including accuracy, sensitivity, and motor irregularity specificity. To address this challenge, this paper proposes HCT, a novel framework that integrates ResNet-based spatial feature extraction from optical flow maps and pose estimation skeletons with multi-head self-attention Transformer encoding for modeling long-range temporal dependencies across multi-frame motion sequences. Unlike conventional single-stream approaches, where spatial and temporal processing remain confined to independent architectures, HCT decouples spatiotemporal feature learning through a cross-modal fusion pipeline, constructing a unified discriminative architecture that captures motor coordination dependencies between motion imaging inputs and multiple diagnostic criteria simultaneously. The convolutional encoder generates diverse joint displacement features, which are consolidated through cross-modal attention fusion into a robust, unified embedding with enhanced generalization and resilience to inter-individual motor variability. Integration within neurodevelopmental assessment frameworks facilitates reliable developmental coordination disorder classification, motor irregularity prediction, and interpretable diagnostic decision support, advancing the accuracy, flexibility, and clinical validity of intelligent motor disorder diagnostic systems. Full article

Background: Achieving remission of type 2 diabetes mellitus (T2DM) after bariatric surgery represents a critical opportunity to reduce long-term diabetes-related complications, including cardiovascular disease, nephropathy, neuropathy, and retinopathy. However, remission rates vary widely across patients, and identifying modifiable clinical and behavioral determinants remains essential for optimizing integrated metabolic care. Objectives: In the current study, we aimed to (1) classify type 2 diabetes mellitus (T2DM) remission status after bariatric surgery through clinical, anthropometric, and behavioral variables at follow-up; (2) identify the main model-based determinants of remission status and explainable machine learning using the preoperative model for baseline risk stratification with surgical candidates. Methods: We performed a retrospective cross-sectional study on 233 patients with T2DM who had bariatric surgery at a tertiary referral center. We made use of two analytical frameworks: a full-feature approach to identify the current remission status in a cross-sectional manner and a preoperative approach to make a temporal classification of the baseline for the first time. We trained and internally assessed 14 machine learning and deep learning classifiers. We evaluated model interpretability using SHAP. Results: In the full-feature cross-sectional classification, the Bottleneck Network performed best (ROC AUC = 0.889). SHAP data identified percentage weight regain, pre- and post-surgical body mass index, HbA1c, and oral hypoglycemic agent use as the dominant model-associated factors. In the restricted preoperative setting, the Extra Trees model achieved an AUC of 0.707, which is a lower level but still represents good baseline risk stratification performance. Conclusions: The findings indicate that remission status after bariatric surgery is both clinical as well as behavioral, but the post-operative or contemporaneously assessed variables should be looked at as classification (as opposed to prediction) models. The preoperative model may be able to be used in risk stratification, but clinical validation and prospective evaluation should be made prior to clinical implementation. Full article

Gut-derived indole metabolites are implicated in the gut–kidney axis, but the factors controlling the intestinal conversion of indole-3-acetic acid (IAA) to skatole remain incompletely defined. We developed a deterministic, hypothesis-generating framework that represents this conversion as a finite-pool allocation process governed by $\mathrm{pH}$-dependent ecological permissiveness, terminal-conversion capacity, precursor availability, spatial progression, and competing loss. The model separates the available-pool scale from a dimensionless integrated conversion exposure, $\Psi$. Across 5400 loss-free scenarios spanning 10 pH profiles and graded metabolic and host-associated constraints, the distal endpoint normalized to the available pool followed the analytically derived relationship $1-\exp \left(-\Psi \right)$. Thus, distinct combinations of mechanistically relevant parameters produced the same normalized distal endpoint, demonstrating that this endpoint alone cannot uniquely identify the underlying mechanism. Competing loss further separated absolute, total-pool-normalized, and conditional outputs, showing that mechanistic interpretation depends on endpoint normalization. Analytical and numerical checks supported internal consistency. The framework was not fitted to biological data, and concentration values were used only as technical scaling references. This biologically unvalidated model generates experimentally testable hypotheses regarding the roles of $\mathrm{pH}$, terminal-conversion capacity, precursor availability, and competing loss in intestinal IAA-to-skatole metabolism; it is not intended to provide physiological or clinical predictions. Full article

Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis is a severe autoimmune neurological disorder characterized by pathogenic antibodies against the NMDAR. A systematic protein profiling approach is warranted to identify biomarkers capable of predicting disease status. An Olink proximity extension assay (PEA) profiled 91 inflammation-related proteins from anti-NMDAR encephalitis patients. Disease severity or prognosis were assessed by CASE score or mRS score at 6-month follow-up. Patients were stratified into distinct molecular clusters using unsupervised clustering. Logistic regression models incorporating selected biomarkers were developed to predict disease severity and prognosis, followed by absolute quantification using ELISA. Patients were classified into four consensus clusters. Clusters 1 and 2 corresponded to the mild group, while Cluster 3 represented the severe group, consistent with CASE score above 6. Cluster 4 showed heterogeneous clinical features. Elevated serum levels of IL-10, IL-6, and SIRT2, as well as increased CSF levels of CXCL10, CXCL11, and MMP10, were positively associated with severe disease. Conversely, several proteins including LTA and CCL11, CCL8, TGFB1, CXCL6 were associated with severe disease or unfavorable 6-month outcomes. A logistic regression model combining serum CXCL6 and CCL11 with CSF MMP10 achieved an area under the curve (AUC) of 0.95 for predicting disease severity. Serum CCL11 alone showed predictive value for 6-month prognosis, with an AUC of 0.79. These findings delineate distinct protein signatures associated with clinical heterogeneity of anti-NMDAR encephalitis. Prediction models incorporating multiple biomarkers may provide an approach for disease severity stratification and prognosis forecast. Full article