Browse Articles
Discover research articles across all indexed journals
Attention re-alignment in multimodal large language models via intermediate-layer guidance
Investigation of multi-performance optimization of banana/bark cloth reinforced epoxy composites using grey relational analysis for automotive interior applications
Abstract This study presents a novel multi-performance optimization of underexplored hybrid banana- and bark-cloth fiber-reinforced epoxy composites for automotive interior applications, using Grey Relational Analysis (GRA). Unlike previous studies that mainly examine banana fiber alone or other conventional natural fiber hybrids, the present work investigates the synergistic combination of banana fiber and bark cloth in an epoxy matrix with 5 wt% aluminium trihydrate (ATH) to simultaneously improve mechanical performance, flame resistance, and moisture-related durability. The composites were produced through compression moulding with varying fiber ratios, and their tensile, flexural, impact, flammability, and water absorption behaviours were systematically evaluated. Among the tested configurations, Sample C7 (40% banana, 5% bark cloth) exhibited the best overall performance, with tensile strength of 28.29 MPa, flexural strength of 37.86 MPa, specific impact energy of 106.92 J/g, and a burning rate of 1.053 mm/s. GRA was used to rank the composite formulations based on multiple response criteria, identifying C7 as the optimal configuration. SEM, EDAX, and FTIR analyses further confirmed improved fibre-matrix interaction, additive retention, and chemical compatibility. The findings demonstrate the potential of banana/bark cloth hybrid composites as sustainable multifunctional materials for automotive interiors.
Retraction Note: Evolutionary analysis of rainstorm momentum and non-stationary variating patterns in response to climatic changes across diverse terrains
Personal trip purpose inference based on dynamic time–age functional zones
The first limb-sparing use of histotripsy for canine osteosarcoma
Abstract Osteosarcoma (OS) is a devastating primary bone cancer in dogs and humans. Histotripsy is a non-invasive ablation technique that mechanically destroys targeted tumor with focused ultrasound and has shown promise in treating small portions of canine OS followed by limb amputation. This study represents the first use of histotripsy to treat large volumes of canine OS, using fractionated treatments, without subsequent surgical resection. Our objective was to evaluate the safety and feasibility of treating large volumes of OS, following clinical outcomes after histotripsy. 9 dogs with suspected OS were treated with 1–5 histotripsy ablations, depending on tumor size, fractionated over 3 to 17 days. Pain, quality of life, and gait were followed over time, and MRI was used for pre-treatment planning and post-treatment monitoring. All dogs tolerated fractionated histotripsy treatments well, and the ablation zone was visible as a non-enhancing area on MRI matching the targeted location. There was a significant reduction in tumor contrast enhancement after completion of all histotripsy treatments and at the end of follow-up. Across all dogs with follow-up, there was a significant increase in pressure on the tumor-bearing limb during walking at the end of follow-up as compared to pre-treatment, indicating decreased pain. 4/6 dogs with follow-up showed a clinically significant decrease in pain. This study shows the promise of histotripsy as a non-invasive limb-sparing treatment for OS and advances the use of fractionated histotripsy treatments to ablate large and complicated tumors.
Analyzing the digital customer journey: a novel framework for sequential behavior modeling
Abstract In recent years, the growing amount of event-log data collected from websites and mobile applications has provided an opportunity to analyoc2vec to cluster physical ze user-behavior patterns in digital customer systems. However, a holistic data-driven approach for analyzing the digital customer journey is lacking. To address this gap, we propose a method for analyzing large amounts of digital event logs to derive actionable insights related to customers’ digital journeys, in order to improve user experience, satisfaction, and loyalty, particularly in high-stakes domains like healthcare. The proposed method identifies the most relevant types of customer journeys via the latent Dirichlet allocation (LDA) algorithm and its hierarchical version (HLDA). The proposed method is executed by generating topics from sequences of events that are transformed into session clusters. We then capture sequential behavioral patterns among and within the clusters and further model them as Markov chains. The results of this method for various applications are presented. We demonstrate the most representative paths and detect customers’ actions with a high probability of potential undesirable outcomes. We utilize a unique real-world dataset of event logs extracted from a mobile application of a large health maintenance organization (HMO). This dataset includes over 120,000 active customers who generated more than 5.5 million events per day. This research makes three main contributions: )i) we propose a novel methodology that integrates established techniques (LDA/hLDA, Markov chain analysis, and leakage/irregularity detection) for analyzing digital customer journeys at various granularity levels to simplify the variability of the customer journey; (ii) we show how topic-based session summarization reduces complexity and supports customer journey’s sequential diagnostics, for example by detecting irregularities of unique behavioral patterns that would otherwise be difficult to detect; and (iii) we demonstrate the framework on a unique real-world HMO dataset that enables the implementation of the method to extract actionable and updated insights important for the digital healthcare domain. Finally, we discuss the emerging role of Artificial Intelligence and Large Language Models in digital customer journey analysis.
A lethal incident during an intergroup encounter in bonobos
Abstract Although neighbouring bonobo communities often display tolerance and cooperation when associating together, aggression is also commonly observed during intergroup encounters. Here, we describe an observation at the Kokolopori research site (DRC) in which a coalition of individuals from the neighbouring Kokoalongo community aggressed an adult female from the Ekalakala community. This happened during an ongoing intergroup encounter that had begun earlier as the two communities travelled and foraged together. Shortly after the attack, the 52-day-old infant of the attacked female was carried by two immatures of Kokoalongo. The infant showed signs of distress before being taken by an adult Kokoalongo female, the mother of the immatures. The infant died the following day, and the adult female continued to carry the corpse for two additional days. This unique case of infant acquisition by unrelated out-group individuals immediately following aggression against the mother underscores that bonobo intergroup encounters can have lethal outcomes. These events highlight the complexity and unpredictability of intergroup dynamics in this species.
An automated model for differential meet in the middle attacks on AndRX ciphers
Japanese cancer survivors have a higher risk of fragility fractures over ten years
Abstract This study aimed to clarify the risk of fragility fractures (FFs) in Japanese cancer survivors compared with individuals without cancer by incorporating time-dependent variables. In a 10-year cohort of 10,330 adults aged 40–69 years, we assessed the risk of FFs (i.e., hip, vertebral compression, and distal radius fractures) associated with cancer status by using time-dependent Cox proportional hazards analyses. The models were adjusted for demographic, lifestyle, and medical factors. The time-dependent variables included cancer status and age. Analyses were performed according to cancer treatment status, number of primary cancers, and cancer site. Among the 10,330 participants (mean age, 56.4 years; females, 6148), 386 experienced FFs. Compared with participants without cancer, a higher risk of overall FFs was observed in participants with cancer (hazard ratio [HR], 1.40; 95% confidence interval [CI], 1.03–1.89). This risk was further elevated among participants with active cancer (HR, 1.71; 95% CI, 1.15–2.55), multiple primary cancers (HR, 1.42; 95% CI, 1.04–1.95), stomach cancer (HR, 1.84; 95% CI, 1.01–3.37), kidney cancer (HR, 3.74; 95% CI, 1.20–11.70), and hematologic cancer (HR, 7.70; 95% CI, 2.45–24.16). These findings highlight the importance of FF prevention strategies in cancer survivorship care in aging Asian populations.
Emergence time of CO2-forced European summer climate trends
Abstract This study investigates the time of emergence (ToE) of European summer climate trends in a warming world. We use a large ensemble of simulations performed with the Kiel Climate Model, in which atmospheric CO 2 levels increase by 1% per year starting from pre-industrial concentration. The ToE for near-surface temperature, soil moisture, and the hydrological cycle highlights a relatively fast (20–40 years) emergence of near-surface temperature trends, while precipitation trends remain within the range of natural variability until the end of the 140-year-long simulation when CO 2 levels quadruple. Soil moisture trends emerge after approximately 30 years in parts of the Mediterranean region, whereas in Western and Central Europe, they only emerge in the west after about 70 years when CO 2 concentrations have doubled. Although many CO 2 -forced climate trends are not emerging over Europe, a comparison of 5,000 pre-industrial and 3,000 4xCO 2 summers reveals statistically significant differences in the distribution of these summer variables: we find that the 1% driest summers are projected to be more extreme in a warmer 4xCO 2 climate compared to the pre-industrial climate. This can have major implications for agriculture, as water shortages may become more severe during these extreme summers.
Evaluating the efficacy of smoke management technologies in laparoscopic sleeve gastrectomy: insights from a prospective, single-centre comparative study
Abstract Surgical power devices generate surgical smoke that may contain infectious components. Various technologies have been developed to improve surgical smoke management, but comparative performance data from human studies are limited. A prospective, single-centre study was performed for evaluating three smoke management technologies - continuous passive filtration (CPF), electrostatic precipitation (ESP), and continuous active filtration (CAF) - during laparoscopic sleeve gastrectomy in 15 bariatric patients. Surgical smoke concentration was monitored by condensation particle counting and single particle light scattering. Efficiency of intraoperative smoke clearance was assessed by the concentration half-life (T 1/2 ). Secondary outcomes included total CO 2 consumption, intraoperative pressure stability, and intraoperative visibility. ESP showed the highest smoke clearance efficiency (T 1/2 = 7.2 s), followed by CAF (18.3 s) and CPF (20.6 s) with significant differences. Total CO 2 consumption was highest for CAF (452.0 L) compared CPF (242.0 L) and ESP (80.1 L). All groups maintained a stable capnoperitoneal pressure and a good/very good intraoperative visibility. Electrostatic precipitation showed the lowest CO 2 consumption and significantly higher smoke particle removal efficiency compared to continuous active/passive filtration. All technologies provided good/very good intraoperative visibility and capnoperitoneal pressure stability.
Diagnostic performance of artificial intelligence for detecting peritoneal and small bowel dissemination in epithelial ovarian cancer using preoperative contrast-enhanced CT imaging
Correction: Estimation of sexual dimorphism of adult human mandibles of South Indian origin using non-metric parameters and machine learning classification algorithms
Quality of TikTok videos on sudden cardiac death varies by video characteristics and health information accuracy
AI-driven data-efficient estimation of partition functions in disordered materials
Abstract This work presents PULSE (Partition function Unsupervised Learning Sampling and Evaluation), a novel generative approach to efficiently estimate the partition function of disordered compounds, without requiring pre-existing training datasets and at a fraction of the computational cost compared to traditional methods. By enabling targeted calculations of atomic-scale properties, PULSE overcomes key limitations of existing sampling techniques, which either require prohibitive computational effort (Monte Carlo) or result in a partial exploration of the configuration space (Special Quasirandom Structures). The approach, based on an inverse variational autoencoder architecture, can also generate representative configuration sets tailored to specific properties, offering a powerful tool for constructing optimized datasets for training interatomic potentials. We demonstrate the capabilities of our method by computing point-defect formation energies and concentrations in uranium-plutonium mixed oxides. Furthermore, we show that the generated configurations provide physical insight into the role of local environments on defect behavior. Beyond this specific application, PULSE is broadly applicable to studying disorder-driven properties in complex materials, including mixed oxides and high-entropy alloys.
Probiotic Enterococcus faecium (M74) as an alternative to antibiotics for controlling necrotic enteritis in broiler chickens
Abstract Necrotic enteritis caused by Clostridium perfringens remains a major challenge in broiler production, particularly under restrictions on antibiotic use. This study evaluated the prophylactic and therapeutic effects of Enterococcus faecium (M74) as a probiotic alternative to antibiotics in broiler chickens experimentally challenged with C. perfringens . Birds were allocated into negative control, positive control, prophylactic, and therapeutic groups, and growth performance, bacterial counts, hematological and biochemical indices, immune responses, intestinal morphometry, and histopathology were assessed. Supplementation with E. faecium (M74) markedly reduced intestinal bacterial load, with C. perfringens and total coliform counts decreased by approximately 88% and 84%, respectively, in the prophylactic group compared with the infected control. Final body weight and body weight gain were improved by approximately 30–31%, while feed conversion ratio was improved by about 25%. Villus height increased by approximately 43%, accompanied by a marked reduction in histopathological lesion scores (up to 82% decrease in degeneration score). Renal and hepatic function markers were significantly ameliorated, and immune parameters were markedly enhanced in treated birds. The prophylactic regimen consistently showed superior efficacy compared with therapeutic application. Although the study was limited to a single probiotic strain and experimental conditions, these findings indicate that E. faecium (M74) is a promising probiotic candidate for mitigating necrotic enteritis in broilers and may represent a viable alternative or complementary strategy to antibiotics.