Search Results (64)

Background/Objectives: Albumin supplementation is widely used for hypoalbuminemia treatment in patients with critical illness, especially those with cirrhosis. However, studies have demonstrated that routine albumin administration is not always advantageous. We examined how albumin supplementation affects survival outcomes in patients with sepsis with hypoalbuminemia. Methods: This study was conducted by researchers in Taiwan using data from the TriNetX research platform, covering the period from 1 April 2014 to 30 April 2024. This platform aggregates real-world data from healthcare organizations worldwide. From this dataset, 1,147,433 patients who developed sepsis and hypoalbuminemia with albumin levels <3.5 g/dL were identified. The study population was stratified into two groups on the basis of whether they received albumin infusion or not. To compare outcomes, hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated between propensity-score-matched patients who did and did not receive albumin supplementation. Subgroup analysis by albumin levels was conducted. Results: Albumin infusion was linked to increased risks of 30-day mortality (HR [95% CI] = 1.800 [1.774–1.827], p < 0.05), shock (HR [95% CI] = 1.436 [1.409–1.465], p < 0.05), septic shock (HR [95% CI] = 1.384 [1.355–1.415], p < 0.05), hypovolemic shock (HR [95% CI] = 1.496 [1.391–1.608], p < 0.05), cardiogenic shock (HR [95% CI] = 1.553 [1.473–1.637], p < 0.05), heart failure (HR [95% CI] = 1.098 [1.080–1.116], p < 0.05), and pulmonary edema (HR [95% CI] = 1.479 [1.438–1.520], p < 0.05). The subgroup analysis by albumin levels revealed a trend of increased mortality risk with albumin supplementation in patients with high baseline albumin levels. Conclusions: Patients with sepsis with hypoalbuminemia who received albumin supplementation exhibited high 30-day mortality rates and increased risks of shock, heart failure, and pulmonary edema compared with those who did not. These findings indicate that routine albumin administration may be linked with unfavorable outcomes in these patients. Full article

Botulinum neurotoxin type A (BoNT-A) is widely used to reduce spasticity in children with cerebral palsy. Despite its therapeutic benefits, incomplete muscle recovery has been observed post-treatment. This study evaluated longitudinal BoNT-A effects on muscle morphology over one year in children with CP (n = 26, mean age: 5.19 years ± 3.26). Three-dimensional freehand ultrasound assessed medial gastrocnemius muscle volume (MV), muscle belly length (ML), cross-sectional area (CSA), and echo intensity (EI) at baseline and at 3, 6, and 12 months post-BoNT-A. Z-score normalization accounted for natural muscle growth. Linear mixed models analyzed muscular changes over time, and repeated-measures ANOVA compared muscle parameters to an age- and severity-matched control group (n = 26, mean age: 4.98 ± 2.15) at one-year follow-up. MV exhibited a declining trend at 3 (p = 0.005), 6 (p = 0.003), and 12 months (p = 0.007), while ML remained unchanged throughout follow-up (p = 0.95). The initially reduced CSA at 6 months (p = 0.0005) recovered at one year, and EI increased only at 3 months post-BoNT-A (p < 0.0001). At one-year follow-up, there was a trend for reduced growth rate (MV/month) (p = 0.035) in the intervention group, whereas the control group exhibited an increased muscle growth (p = 0.029). These findings suggest distinct recovery timelines for CSA and ML, which may explain the incomplete MV recovery and highlight substantial interindividual variation in recovery processes. Full article

Proactively hunting known attack behaviors within system logs, termed threat hunting, is gaining traction in cybersecurity. Existing methods typically rely on constructing a query graph representing known attack patterns and identifying it as a subgraph within a system-wide provenance graph. However, the large scale and redundancy of provenance data lead to poor matching efficiency and high false-positive rates. To address these issues, this paper introduces SGMNet, a supervised seeded graph-matching network designed for efficient and accurate threat hunting. By selecting indicators of compromise (IOCs) as initial seed nodes, SGMNet extracts compact subgraphs from large-scale provenance graphs, significantly reducing graph size and complexity. It then learns adaptive node-expansion strategies to capture relevant context while suppressing irrelevant noise. Experiments on four real-world system log datasets demonstrate that SGMNet achieves a runtime reduction of over 60% compared to baseline methods, while reducing false positives by 35.2% on average. These results validate that SGMNet not only improves computational efficiency but also enhances detection precision, making it well suited for real-time threat hunting in large-scale environments. Full article

While deep learning-based image matching methods excel at extracting high-level semantic features from remote sensing data, their performance degrades significantly under cross-daylight conditions and wide-baseline geometric distortions, particularly in multi-source street-view scenarios. This paper presents a novel illumination-invariant framework that synergistically integrates geometric topology and semantic consistency to achieve robust multi-view matching for cross-daylight urban perception. We first design a self-supervised learning paradigm to extract illumination-agnostic features by jointly optimizing local descriptors and global geometric structures across multi-view images. To address extreme perspective variations, a homography-aware transformation module is introduced to stabilize feature representation under large viewpoint changes. Leveraging a graph neural network with hierarchical attention mechanisms, our method dynamically aggregates contextual information from both local keypoints and semantic topology graphs, enabling precise matching in occluded regions and repetitive-textured urban scenes. A dual-branch learning strategy further refines similarity metrics through supervised patch alignment and unsupervised spatial consistency constraints derived from Delaunay triangulation. Finally, a topology-guided multi-plane expansion mechanism propagates initial matches by exploiting the inherent structural regularity of street scenes, effectively suppressing mismatches while expanding coverage. Extensive experiments demonstrate that our framework outperforms state-of-the-art methods, achieving a 6.4% improvement in matching accuracy and a 30.5% reduction in mismatches under cross-daylight conditions. These advancements establish a new benchmark for reliable multi-source image retrieval and localization in dynamic urban environments, with direct applications in autonomous driving systems and large-scale 3D city reconstruction. Full article

Objectives: To evaluate the clinical heterogeneity of sarcomas by examining associations between histological subtypes, metastatic patterns, treatment modalities, and survival outcomes. Methods: We analyzed data from 97,062 adult patients diagnosed with sarcoma between 2000 and 2020, using the Surveillance, Epidemiology, and End Results (SEER) database. Fourteen histological subtypes were included. Propensity score matching (PSM) was employed to adjust for baseline differences, and Cox proportional hazards models were used to identify prognostic variables. Results: The most prevalent subtypes were sarcoma not otherwise specified (31.9%), leiomyosarcoma (17.1%), and liposarcoma (13.9%). Metastatic patterns differed significantly by subtype; liver metastases were most common in sarcomas with small blue round cell (SBRC) features (8.9%) and stromal sarcoma (6.1%), while lung metastases were frequently observed in Ewing sarcoma (10.0%) and rhabdomyosarcoma (9.7%). Median overall survival (mOS) varied widely, ranging from 234 months in chondrosarcoma to 16–20 months in rhabdomyosarcoma and SBRC sarcoma. Overall, patients with primary sarcoma had significantly better survival than those with treatment-related disease (119.0 vs. 45.0 months, p < 0.0001), with this trend consistent across most subtypes. Treatment responses were subtype- and size-dependent. For instance, surgery plus radiotherapy improved outcomes in giant cell sarcoma regardless of tumor size, whereas chemotherapy provided benefit only in tumors larger than 5 cm. Combined surgery and radiotherapy offered additional survival benefit in select subtypes, including chordoma, leiomyosarcoma (>5 cm), and synovial sarcoma (<5 cm). Conclusions: Sarcomas exhibit substantial clinical and prognostic heterogeneity across histological subtypes. These findings underscore the importance of subtype-specific, individualized treatment strategies in optimizing patient outcomes. Full article

Data cleaning remains one of the most time-consuming and critical steps in modern data science, directly influencing the reliability and accuracy of downstream analytics. In this paper, we present a comprehensive evaluation of five widely used data cleaning tools—OpenRefine, Dedupe, Great Expectations, TidyData (PyJanitor), and a baseline Pandas pipeline—applied to large-scale, messy datasets spanning three domains (healthcare, finance, and industrial telemetry). We benchmark each tool on dataset sizes ranging from 1 million to 100 million records, measuring execution time, memory usage, error detection accuracy, and scalability under increasing data volumes. Additionally, we assess qualitative aspects such as usability and ease of integration, reflecting real-world adoption concerns. We incorporate recent findings on parallelized data cleaning and highlight how domain-specific anomalies (e.g., negative amounts in finance, sensor corruption in industrial telemetry) can significantly impact tool choice. Our findings reveal that no single solution excels across all metrics; while Dedupe provides robust duplicate detection and Great Expectations offers in-depth rule-based validation, tools like TidyData and baseline Pandas pipelines demonstrate strong scalability and flexibility under chunk-based ingestion. The choice of tool ultimately depends on domain-specific requirements (e.g., approximate matching in finance and strict auditing in healthcare) and the magnitude of available computational resources. By highlighting each framework’s strengths and limitations, this study offers data practitioners clear, evidence-driven guidance for selecting and combining tools to tackle large-scale data cleaning challenges. Full article

Graph neural networks (GNNs) are widely used for graph-structured data. However, GNNs are vulnerable to membership inference attacks (MIAs) in graph classification tasks, which determine whether a graph was in the training set, risking the leakage of sensitive data. Existing MIAs rely on prediction probability vectors, but they become ineffective when only prediction labels are available. We propose a Graph-level Label-Only Membership Inference Attack (GLO-MIA), which is based on the intuition that the target model’s predictions on training data are more stable than those on testing data. GLO-MIA generates a set of perturbed graphs for the target graph by adding perturbations to its effective features and queries the target model with the perturbed graphs to obtain their prediction labels, which are then used to calculate the robustness score of the target graph. Finally, by comparing the robustness score with a predefined threshold, the membership of the target graph can be inferred correctly with high probability. Experimental evaluations on three datasets and four GNN models demonstrate that GLO-MIA achieves an attack accuracy of up to 0.825, outperforming baseline work by 8.5% and closely matching the performance of probability-based MIAs, even with only prediction labels. Full article

Background and Objectives: Scleral lenses are widely used for visual rehabilitation in keratoconus patients, but their long-term effects on corneal tomography remain unclear. This study aims to evaluate the impact of 12-month scleral lens wear on corneal tomography in keratoconus patients through a case-controlled design. Materials and Methods: This retrospective study included 220 keratoconus patients, of whom 10 eyes were treated with SoClear (Brighten Optix Corporation, Taipei, Taiwan) mini-scleral lenses for over one year (SL group). A control group of 14 eyes was matched using Mahalanobis distance matching based on anterior maximum keratometry (K_max) and age. Both groups were evaluated at baseline and 12 months. Corneal tomography was assessed using the Pentacam HR (Oculus, Wetzlar, Germany), analyzing parameters such as anterior and posterior corneal curvature, thinnest corneal thickness (TCT), and higher-order aberrations. Generalized estimating equations (GEEs) were employed to assess the time-by-treatment effect between the two groups. Results: The SL group included 10 eyes from eight patients (seven males, one female; mean age 30.40 ± 6.52 years), while the control group included 14 eyes from 11 patients (three males, wight females; mean age 27.43 ± 8.11 years). Best corrected visual acuity with spectacles improved significantly with scleral lenses (p = 0.011) and remained stable (p = 0.044) at 12 months. Significant interaction effects were found in Ambrósio relational thickness (p = 0.006), posterior radius curvature (p = 0.047), posterior mean keratometry (p = 0.019), posterior flat keratometry (p = 0.023), and thinnest corneal thickness angle (p = 0.023); the SL group demonstrated less progression in these parameters compared to the control group. Conclusions: This case-controlled study highlights the 12-month impact of scleral lenses on keratoconus, showing improved visual acuity compared to spectacles, stabilized posterior corneal curvature, and maintained corneal thickness. Further prospective studies with larger cohorts are needed to assess scleral lens effect on keratoconus progression. Full article

This study introduces a visual SLAM real-time system designed for small indoor environments. The system demonstrates resilience against significant motion clutter and supports wide-baseline loop closing, re-localization, and automatic initialization. Leveraging state-of-the-art algorithms, the approach presented in this article utilizes adapted Oriented FAST and Rotated BRIEF features for tracking, mapping, re-localization, and loop closing. In addition, the research uses an adaptive threshold to find putative feature matches that provide efficient map initialization and accurate tracking. The assignment is to process visual information from the camera of a DJI Tello drone for the construction of an indoor map and the estimation of the trajectory of the camera. In a ’survival of the fittest’ style, the algorithms selectively pick adaptive points and keyframes for reconstruction. This leads to robustness and a concise traceable map that develops as scene content emerges, making lifelong operation possible. The results give an improvement in the RMSE for the adaptive ORB algorithm and the adaptive threshold (3.280). However, the standard ORB algorithm failed to achieve the mapping process. Full article

Background/Objectives: Proton pump inhibitors (PPIs) are widely used for acid-related gastrointestinal disorders, but their potential association with lung cancer risk and mortality remains underexplored and debated. This study sought to investigate the association between PPI use and lung cancer likelihood and mortality, focusing on the impact of PPI exposure history and duration. Methods: This study utilized data from 6795 lung cancer patients, 27,180 matched controls, and 4257 deceased and 2538 surviving lung cancer patients from the Korean National Health Insurance Service’s Health Screening Cohort (2002–2019). Propensity score overlap weighting and logistic regression models were applied to assess the correlations between PPI usage history and duration with lung cancer risk and mortality, while standardized differences ensured balanced baseline characteristics. Results: Overall, PPI use was modestly associated, with a 19% increased likelihood of lung cancer occurrence (95% confidence intervals (CI): 1.12–1.26). Interestingly, prolonged PPI use (≥30 days) was linked to a 13% reduction in lung cancer incidence (95% CI: 0.80–0.94), particularly in subgroups such as older adults (≥70 years), individuals with gastroesophageal reflux disease (GERD) or hypertension, and those with low alcohol consumption. Conversely, overall PPI usage was linked with a 36% increased mortality likelihood among lung cancer patients (95% CI: 1.20–1.55), with prolonged use further correlating with a 27% higher mortality risk (95% CI: 1.05–1.53), especially in high-risk subgroups, including smokers, underweight individuals, and those with hypercholesterolemia or GERD. Conclusions: These findings may suggest a complex and context-dependent relationship between PPI use and lung cancer outcomes, emphasizing the need for individualized risk assessments and careful prescribing practices. Full article

Background/Objectives: Aciclovir is a widely used antiviral agent. Since aciclovir is primarily eliminated through the kidneys, maintaining renal function is crucial to avoid toxicity. Although mitigating strategies are introduced in the standard of care, nephrotoxicity is still a major concern during treatment, especially for critically ill intensive care unit (ICU) patients. Therefore, risk factors for the development of nephrotoxicity during aciclovir therapy should be addressed. This study aimed to evaluate if aciclovir in combination with therapeutic drug monitoring (TDM) and additional nephrotoxicity-mitigating strategies is associated with a decrease in renal function in critically ill ICU patients. Methods: In a cohort of ICU patients with or without intravenous aciclovir treatment (including standard of care mitigating strategies) propensity score matching was applied to balance baseline characteristics between aciclovir-treated and untreated groups. Aciclovir was monitored by measuring serum levels and the dose was adjusted when needed. Renal function was primarily assessed through serum creatinine. Univariate and multivariate regression analyses were used to identify risk factors for nephrotoxicity during ICU admission. Results: After propensity score matching, the study included 518 ICU patients, of whom 259 received aciclovir. Aciclovir was not associated with a significant decrease in renal function during admission. In fact, renal function appeared to improve in the aciclovir-treated group compared to the control group (beta-coefficient: −14.5 (95% confidence interval: −28.3 to −0.68), p = 0.04). Median aciclovir concentrations remained within the exploratory therapeutic range. Conclusions: Aciclovir therapy, at least when appropriately monitored, does not independently induce nephrotoxicity in critically ill ICU patients. TDM may further enhance safety by preventing supratherapeutic drug exposures. The results are significant as they provide evidence supporting the safe use of aciclovir in a vulnerable patient population. Future studies should focus on establishing therapeutic and toxic concentration thresholds for aciclovir and assessing the clinical utility of TDM in this context. Full article

Existing person re-identification methods utilizing CLIP (Contrastive Language-Image Pre-training) mostly suffer from coarse-grained alignment issues. This is primarily due to the original design intention of the CLIP model, which aims at broad and global alignment between images and texts to support a wide range of image–text matching tasks. However, in the specific domain of person re-identification, local features and fine-grained information are equally important in addition to global features. This paper proposes an innovative modal fusion approach, aiming to precisely locate the most prominent pedestrian information in images by combining visual features extracted by the ResNet-50 model with text representations generated by a text encoder. This method leverages the cross-attention mechanism of the Transformer Decoder to enable text features to dynamically guide visual features, enhancing the ability to identify and locate the target pedestrian. Experiments conducted on four public datasets, namely MSMT17, Market1501, DukeMTMC, and Occluded-Duke, demonstrate that our method outperforms the baseline network by 5.4%, 2.7%, 2.6%, and 9.2% in mAP, and by 4.3%, 1.7%, 2.7%, and 11.8% in Rank-1, respectively. This method exhibits excellent performance and provides new research insights for the task of person re-identification. Full article

Remote sensing image classification has achieved remarkable success in environmental monitoring and urban planning using deep neural networks (DNNs). However, the performance of these models is significantly impacted by domain shifts due to seasonal changes, varying atmospheric conditions, and different geographical locations. Existing solutions, including rehearsal-based and prompt-based methods, face limitations such as data privacy concerns, high computational overhead, and unreliable feature embeddings due to domain gaps. To address these challenges, we propose DACL (dual-pool architecture with contrastive learning), a novel framework for domain incremental learning in remote sensing image classification. DACL introduces three key components: (1) a dual-pool architecture comprising a prompt pool for domain-specific tokens and an adapter pool for feature adaptation, enabling efficient domain-specific feature extraction; (2) a dual loss mechanism that combines image-attracting loss and text-separating loss to enhance intra-domain feature discrimination while maintaining clear class boundaries; and (3) a K-means-based domain selector that efficiently matches unknown domain features with existing domain representations using cosine similarity. Our approach eliminates the need for storing historical data while maintaining minimal computational overhead. Extensive experiments on six widely used datasets demonstrate that DACL consistently outperforms state-of-the-art methods in domain incremental learning for remote sensing image classification scenarios, achieving an average accuracy improvement of 4.07% over the best baseline method. Full article

Thermal ablation has been widely used for patients with small colorectal liver metastases (CRLMs), even for resectable cases; however, solid evidence has been scarce. (1) Thermal ablation versus liver resection. Some propensity-score matching studies using patients with balanced baseline characteristics have confirmed less invasiveness and the comparable survival benefits of thermal ablation to liver resection. A more recent pivotal randomized controlled trial comparing thermal ablation and liver resection was presented during the American Society of Clinical Oncology 2024 meeting. Diameter ≤ 3 cm, ten or fewer resectable and ablatable CRLMs were assigned to thermal ablation or liver resection. No differences were observed in the overall survival and local and distant progression-free survival with less morbidity. (2) Combination of thermal ablation and liver resection. Four matching studies demonstrated comparable data between the combination and liver resection alone groups in the long-term survival and recurrence rates without increasing the postoperative complication rates. The selection of the two approaches depends primarily on the number, size, and location of the CRLMs. (3) Chemotherapy in combination with thermal ablation. A propensity-score matching study comparing thermal ablation ± neoadjuvant chemotherapy was conducted. The addition of neoadjuvant chemotherapy was an independent predictive factor for good progression-free survival without increasing morbidity. Two randomized controlled trials demonstrated that additional thermal ablation to systemic chemotherapy can improve the overall survival for initially unresectable CRLMs. (4) Conclusions. Thermal ablation can provide survival benefits for patients with CRLMs in various situations, keeping adequate indications. Full article

Raman spectroscopy provides detailed information about the molecular composition of a sample. The classical identification of components in a multi-component sample typically involves comparing the preprocessed spectrum with a known reference stored in a database using various spectral matching or machine-learning techniques or relies on universal models based on a two-step analysis including first, the component identification, and then the decomposition of the mixed signal. However, although large databases and universal models cover a wide range of target materials, they may be not optimized to the variability required in a specific application. In this study, we propose a single-step method using deep learning (DL) modeling to decompose a simulated mixture of real measurements of Raman scattering into relevant individual components regardless of noise, baseline and the number of components involved and quantify their ratios. We hypothesize that training a custom DL model for applications with a fixed set of expected components may yield better results than applying a universal quantification model. To test this hypothesis, we simulated 12,000 Raman spectra by assigning random ratios to each component spectrum within a library containing 13 measured spectra of organic solvent samples. One of the DL methods, a fully connected network (FCN), was designed to work on the raw spectra directly and output the contribution of each component of the library to the input spectrum in form of a component ratio. The developed model was evaluated on 3600 testing spectra, which were simulated similarly to the training dataset. The average component identification accuracy of the FCN was 99.7%, which was significantly higher than that of the universal custom trained DeepRaman model, which was 83.1%. The average mean absolute error for component ratio quantification was 0.000562, over one order of magnitude smaller than that of a well-established non-negative elastic net (NN-EN), which was 0.00677. The predicted non-zero ratio values were further used for component identification. Under the assumption that the components of a mixture are from a fixed library, the proposed method preprocesses and decomposes the raw data in a single step, quantifying every component in a multicomponent mixture, accurately. Notably, the single-step FCN approach has not been implemented in the previously reported DL studies. Full article