Search Results (1,245)

Manufacturing companies face significant challenges in leveraging artificial intelligence for equipment management due to high infrastructure costs and limited availability of labeled data for failures. While most manufacturing AI applications focus on structured sensor data, vast amounts of unstructured textual information containing valuable maintenance knowledge remain underutilized. This study presents a practical generative AI-based framework for structured information extraction that automatically converts unstructured equipment maintenance texts into predefined semantic fields to support predictive maintenance in manufacturing environments. We adopted and evaluated three representative generative models—Bidirectional and Auto-Regressive Transformers (BART) with KoBART, Text-to-Text Transfer Transformer (T5) with pko-t5-base, and the large language model Qwen—to generate structured outputs by extracting three predefined fields: failed components, failure types, and corrective actions. The framework enables the structuring of equipment management text data from Manufacturing Execution Systems (MES) to build predictive maintenance support systems. We validated the approach using a large-scale MES dataset consisting of 29,736 equipment maintenance records from a major automotive parts manufacturer, from which curated subsets were used for model training and evaluation. Our methodology employs Generative Pre-trained Transformer 4 (GPT-4) for initial dataset construction, followed by domain expert validation to ensure data quality. The trained models achieved promising performance when evaluated using extraction-aligned metrics, including exact match (EM) and token-level precision, recall, and F1-score, which directly assess field-level extraction correctness. ROUGE scores are additionally reported as a supplementary indicator of lexical overlap. Among the evaluated models, Qwen consistently outperformed BART and T5 across all extracted fields. The structured outputs are further processed through domain-specific dictionaries and regular expressions to create a comprehensive analytical database supporting predictive maintenance strategies. We implemented a web-based analytics platform enabling time-series analysis, correlation analysis, frequency analysis, and anomaly detection for equipment maintenance optimization. The proposed system converts tacit knowledge embedded in maintenance texts into explicit, actionable insights without requiring additional sensor installations or infrastructure investments. This research contributes to the manufacturing AI field by demonstrating a comprehensive application of generative language models to equipment maintenance text analysis, providing a cost-effective approach for digital transformation in manufacturing environments. The framework’s scalability and cloud-based deployment model present significant opportunities for widespread adoption in the manufacturing sector, supporting the transition from reactive to predictive maintenance strategies. Full article

Mass M (kg) and peak intrusion L (mm) are jointly minimized for a CFRP-enabled battery pack enclosure under the GB 38031-2025 −40° side-pole extrusion condition. A 50-run explicit FE design of experiments is conducted and deterministically partitioned into 37/5/5/3 for initial training, two sequential enrichment batches, and an independent hold-out test. Bayesian additive regression trees are trained as the primary surrogates for M, L, and Stress, and stress acceptability is enforced through a probability-of-feasibility (PoF) gate anchored to a baseline-scaled cap, σ_lim = 1.2 σ_base = 410.4 MPa. NSGA-II performed on the feasible surrogate landscape yields a bimodal feasible non-dominated set. The two branches correspond to two discrete levels of a key thickness variable x₄: a low-mass regime (n = 106) with M = 100.61–104.81 kg and L = 5.430–5.516 mm at x₄ ≈ 5.60 mm, and a stiffer regime (n = 94) with M = 110.69–115.08 kg and L = 5.362–5.430 mm at x₄ ≈ 8.00 mm. PoF screening eliminates part of the intermediate region where feasibility confidence is insufficient. Independent FE reruns further indicate that the PoF gate reduces deterministic misclassification near the stress boundary (e.g., one near-threshold candidate exceeds σ_lim, whereas others satisfy the cap with margin). Overall, the proposed workflow offers a traceable lightweighting route under extreme-cold uncertainty within a constrained FE budget. Full article

Background: Tumor spread through air spaces (STAS) has been proposed as a histopathological marker of aggressive tumor biology in adenocarcinoma of the lung (ADCL). Its independent prognostic significance and clinical implications regarding surgical strategy remain controversial. This study evaluated clinicopathological correlates and the prognostic impact of STAS in a homogeneous cohort of resected ADCL. Methods: We retrospectively analyzed 100 patients with primary ADCL resected between 2009 and 2018. STAS was classified as absent, low (1–4 clusters), or high (≥5) by an experienced pathologist. Associations between STAS and clinical, surgical, and pathological variables were tested with univariate analyses and multivariable logistic regression. Overall survival (OS) was evaluated using Kaplan–Meier and Cox regression. Results: STAS was present in 46% of tumors and was significantly associated with a higher pathological N category (pN0-pN3; p = 0.005), more advanced UICC stage (p = 0.049), lymphovascular invasion (LVI; p = 0.008), and perineural invasion (PnI; p = 0.012). In univariate survival analysis, patients with STAS had shorter OS than patients without STAS (p = 0.047). After limited resection, OS did not differ (p = 0.864), whereas after radical anatomical resection, patients with STAS showed reduced OS (p = 0.034). In multivariable Cox regression analysis, STAS did not retain independent prognostic significance. Conclusions: STAS is frequent in resected ADCL and correlates with adverse pathological features and reduced OS in univariate models. In multivariate analysis, STAS did not emerge as an independent prognostic factor. These findings support the interpretation of STAS as a marker of aggressive tumor biology rather than an independent determinant of prognosis or surgical decision-making. Full article

The global lithium supply balance has been under pressure since the recent increase in demand for electric vehicles. Conventional techniques for lithium extraction from natural resources are solar evaporation and hard-rock mining, which both have their limitations in view of sustainability. The question arises whether these methods will suffice for a responsible supply to provide the necessary materials for the emerging green economy. While new technologies for the valorization of lithium from unconventional resources like geothermal brines, salt lakes and seawater are in the pipeline, they are yet to be proven on an industrial scale. Membrane technology, ion-exchange adsorption and electrochemical methods are the current focus of several players in the pilot stage of their announced lithium carbonate or hydroxide production process. These technologies have various advantages and disadvantages in terms of energy consumption, selectivity and process costs, and the optimal choice remains dependent on local factors such as brine composition, energy availability and reagent cost. There are currently several DLE projects in the pilot phase, which is a significant step towards more sustainable lithium supply. Proving the economic and technical viability of these methods for extracting lithium from unconventional sources would increase the amount of globally proven reserves while diversifying and de-risking the supply chain, which is currently heavily dominated by a small number of countries. Full article

Piglets are highly susceptible to iron deficiency. This randomized clinical trial evaluated the effectiveness of four iron dosing schemes in preventing anemia. Two herds with different farrowing management systems were included. In each herd, 40 litters (6 piglets/litter) were selected on day 3 of age. A 2 × 2 factorial design was applied, combining two intramuscular iron dextran injection schemes [37.5 mg Fe/kg (low injection; LI) or 150 mg Fe/kg (high injection; HI)] with two oral ferrous sulphate feed supplementation schemes [125 mg Fe/kg (low feed; LF) or 200 mg Fe/kg (high feed; HF)]. Blood samples were collected at 4 and 20 days of age, and piglets were weighed at 3 and 20 days. Data were analyzed using linear mixed models, with significance set at p < 0.05. In Herd A, HI-LF piglets showed increased body weight, whereas no growth differences were observed in Herd B. Creep-feed intake did not differ between treatments. HI consistently improved red-cell indices in Herd A, while in Herd B LI piglets initially showed higher values at day 4, but HI piglets surpassed them by day 20. Leukocyte responses were limited. High-dose iron injections were effective in preventing anemia, while oral supplementation had minimal impact. Full article

Zirconia is widely used in full-arch restorations due to its strength and aesthetics, but failures can still affect its performance in clinical practice. In this report, a full-arch tooth-supported zirconia bridge fractured prematurely (eleven months), encouraging an investigation into its design and failure mechanisms. STL files obtained from the dental laboratory revealed regions of reduced framework thickness, falling below the manufacturer’s recommendations. Fractographic analysis of the fractured pieces indicated a multifactorial failure pattern. Notable features included a thick cement layer, surface damage likely caused by the CAM bur during milling, and occlusal wear affecting the glazed surface. Crack propagation was observed in an occlusal-to-cervical direction. While no single factor could be definitively identified as the primary cause, the failure is attributed to the combined effect of insufficient design, surface damage, and biomechanical overload. Importantly, most such factors are not visible before failure, raising questions about the proper evaluation of zirconia-based restorations prior to their cementation. Full article

Background: Durable bonding to zirconia remains difficult because its chemically inert surface resists acid etching. Additive manufacturing (AM) enables controlled surface morphology, which may enhance micromechanical retention without additional treatments. Methods: Zirconia specimens with three AM-derived surface designs—(1) concave–convex hemispherical patterns, (2) concave hemispherical patterns, and (3) as-printed surfaces—were fabricated using a slurry-based 3D printing system and sintered at 1500 °C. Zirconia specimens fabricated by subtractive manufacturing using CAD/CAM systems, polished with 15 µm diamond lapping film and with or without subsequent alumina sandblasting, served as controls. Surface morphology was analyzed by FE-SEM, and shear bond strength (SBS) was tested after cementation with a resin-based luting agent. Results: SEM revealed regular layered textures and designed hemispherical structures (~300 µm) in AM specimens, along with step-like irregularities (~40 µm) at layer boundaries. The concave–convex AM group showed significantly higher SBS than both sandblasted and polished subtractive-manufactured zirconia (p < 0.05). Vertically printed specimens demonstrated greater bonding strength than those printed parallel to the bonding surface, indicating that build orientation affects resin infiltration and interlocking. Conclusion: AM-derived zirconia surfaces can provide superior and reproducible micromechanical retention compared with conventional treatments. Further optimization of printing parameters and evaluation of long-term durability are needed for clinical application. Full article

Background: ExPEC9V is a 9-valent vaccine candidate designed to prevent invasive Escherichia coli disease, a life-threatening condition occurring when extraintestinal pathogenic E. coli (ExPEC) invade sterile sites. We evaluated immunogenicity and safety when ExPEC9V was co-administered with high-dose (HD) quadrivalent seasonal influenza vaccine. Methods: This Phase 3, double-blind, placebo-controlled study (NCT06134804) randomized 959 adults (≥65 years) to receive co-administration of ExPEC9V and HD quadrivalent seasonal influenza vaccine (CoAd) or each vaccine alone, 29 days apart (Control). Co-primary objectives were non-inferiority of co-administration versus separate administration following predefined criteria based on influenza strain-specific hemagglutination inhibition (HAI) antibody titers and ExPEC9V O-serotype binding antibody levels (multiplex electrochemiluminescence-based immunoassay), 29 days post vaccination. Reactogenicity and safety were assessed. Results: Co-administration of ExPEC9V with HD influenza vaccine demonstrated non-inferiority (upper bound of 2-sided 95% confidence interval [CI] < 1.5 for HAI geometric mean ratio [Control/CoAd]) for all influenza strains. Non-inferiority for ExPEC9V O-serotype antibody levels was not demonstrated (upper bound 95% CI > 1.5). One of nine serotypes met the non-inferiority criterion; eight did not, with four narrowly failing to meet the non-inferiority criterion. ExPEC9V immunogenicity was similar regardless of urinary tract infection history. ExPEC9V was safe and well tolerated, with no serious adverse events related to ExPEC9V. Reactogenicity rate was higher with co-administration. Conclusions: Co-administration of ExPEC9V with HD influenza vaccine met non-inferiority criteria of humoral immune responses for influenza antigens, but not for ExPEC9V O-serotype antigens. ExPEC9V, administered alone or with HD influenza vaccine, was safe and well tolerated, with an acceptable reactogenicity profile. Full article

Spiritually integrated group therapy aims to support coping, meaning-making, and existential recovery in patients receiving psychiatric care. SPIRIT (Spiritual Psychotherapy for Inpatient, Residential and Intensive Treatment) is a structured, flexible protocol implementing this approach into intensive or residential settings. This study examines (1) the impact of SPIRIT on patients’ lives and (2) their needs in terms of aftercare to determine whether and how its benefits can be sustained long term. Data were collected from multiple sources: patient evaluation forms (n = 118); in-depth interviews with patients (n = 19) and caregivers providing the therapy (n = 8); and two focus groups with both caregivers and patients. Transcripts were analyzed using qualitative content analysis. Results indicate that for most participants, the therapy positively impacted their lives through increased awareness or behavioral change, highlighting the relevance of maintaining these insights after group therapy, either at home or within the treatment setting. We recommend broader training of mental health professionals, and the introduction of programs like these to the entire care team to ensure awareness and support. A ‘dual motor model’ is proposed. Addressing religious, spiritual, and meaning-related themes in ongoing therapy and the psychosocial and pastoral support network can support recovery both by reducing symptoms and by fostering a health-promoting context. Full article

Reliable probabilistic wind speed forecasts are essential for integrating renewable energy into power grids and managing operational uncertainty. This study compares Quantile Regression Forests (QRF), Bayesian Additive Regression Trees (BART), and Gaussian Process Regression (GPR) under at-site and regional pooled frameworks using 21 years (2000–2020) of daily wind data from eleven stations in New South Wales and Queensland, Australia. Models are evaluated via strict year-based holdout validation across seven metrics: RMSE, MAE, R², bias, correlation, coverage, and Continuous Ranked Probability Score (CRPS). Regional QRF achieves exceptional point forecast stability with minimal RMSE increase but suffers persistent under-coverage, rendering probabilistic bounds unreliable. BART attains near-nominal coverage at individual sites but experiences catastrophic calibration collapse under regional pooling, driven by fixed noise priors inadequate for spatially heterogeneous data. In contrast, GPR maintains robust probabilistic skill regionally despite larger point forecast RMSE penalties, achieving the lowest overall CRPS and near-nominal coverage through kernel-based variance inflation. Variable importance analysis identifies surface pressure and minimum temperature as dominant predictors (60–80%), with spatial covariates critical for regional differentiation. Operationally, regional QRF is prioritised for point accuracy, regional GPR for calibrated probabilistic forecasts in risk-sensitive applications, and at-site BART when local data suffice. These findings show that Bayesian machine learning methods can effectively navigate the trade-off between local specificity and regional pooling, a challenge common to wind forecasting in diverse terrain globally. The methodology and insights are transferable to other heterogeneous regions, providing guidance for probabilistic wind forecasting and renewable energy grid integration. Full article

Although student evaluations of teaching (SET) in higher education have recently gained considerable interest, little is known about whether course level influences these evaluations. Accordingly, two datasets, obtained from a large public university in the U.S. Midwest, were analyzed to investigate whether course level makes a difference in SET. The first dataset included 25,306 evaluations across eight course levels collected using the SET questionnaire. A one-way multivariate analysis of variance (MANOVA), followed by univariate analyses of variance (ANOVAs), was conducted to test whether course level makes a difference in SET scores. To cross-validate the results and ensure generalizability, the same analyses were conducted using a second, smaller dataset (N = 444). MANOVA results revealed a statistically significant effect for course level on the combined SET dimensions across both campuses. All univariate ANOVAs were also significant across both campuses. Follow-up post hoc comparisons, with level 1 as the reference group, were statistically significant, especially for level 8. Overall, these results underscore the importance of accounting for course level when interpreting SET and encourage researchers to include key covariates (e.g., class size, discipline, instructor experience, and student composition) to identify the factors driving course-level differences. Full article

Background: IDH1-mutant oligodendroglioma and astrocytoma differ not only in growth rate but also in growth pattern. Oligodendrogliomas tend to infiltrate white matter tracts, whereas astrocytomas more often displace them. Such difference could lead to different cognitive outcomes. This study examined differences in executive functioning before and up to one year after surgery between patients with IDH1-mutant astrocytoma and oligodendroglioma. Methods: Patients with WHO grade 2–3 IDH1-mutant oligodendroglioma (1p19q-codeleted) or astrocytoma were included. Cognition was assessed preoperatively, and at 3 and 12 months postoperatively using standardized computerized and paper-and-pencil tests. Groups were compared on demographics, tumor characteristics, surgical modality, extent of resection, adjuvant treatment, and baseline cognition. Longitudinal mixed models were performed to investigate differences in performances over time for the total sample and stratified by surgical approach (awake vs. asleep). Results: 162 patients (67 oligodendroglioma, 95 astrocytoma) were included. Oligodendroglioma patients were older, with more frontal and fewer temporal tumors. Oligodendroglioma patients showed a greater impairment prevalence on a measure of inhibition before surgery. In the awake surgery group, no longitudinal differences were found between diagnoses. In the asleep surgery group, astrocytoma patients remained stable while oligodendroglioma patients declined on a measure of cognitive flexibility, with performance at 3 and 12 months significantly lower than at baseline. Conclusions: Specific aspects of executive functioning in IDH1-mutant gliomas may differ by subtype. Oligodendroglioma patients showed postoperative decline in cognitive flexibility that did not recover to baseline level, particularly in case of surgery under general anesthesia. These results highlight the potential relevance of tumor subtype and surgical approach in limiting cognitive risks after glioma surgery. Full article

Incorporating sensors such as microelectromechanical system (MEMS)-based inertial measurement units (IMUs) and strain gauges into aircraft structures has the potential to complement ground vibration testing results and improve the tracking of structural modes and wing shape in flight, as well as structural health monitoring. This study evaluates the feasibility and accuracy of employing MEMS accelerometers and gyroscopes together with strain gauges to estimate the structural modes of an aircraft. For this purpose, a ground vibration test was carried out on a 1:3 scaled Diana 2 glider model from which the displacement, rotation, and strain modes were estimated. The estimated modal parameters were compared with traditional piezoelectric accelerometer results and Finite Element Method model predictions. The results showed that the modal frequencies, damping ratios, and mode shapes estimated using MEMS IMUs and strain gauges closely matched the reference accelerometer estimates. Furthermore, the combination of displacement, rotation, and strain mode shapes allowed for greater insight into the structural dynamics. The exploratory use of gyroscopes for aircraft GVT allowed the structural torsion to be captured directly, thereby potentially simplifying future GVT setups by eliminating the need for placing accelerometers in pairs across the structure. Full article

Background: the primary challenge with implant infections is the formation of biofilm, which harbors dormant bacteria that reduce the effectiveness of antibiotics and amplify antibiotic resistance, exacerbating the global antimicrobial resistance crisis. A potential novel treatment strategy is radioimmunotherapy, which uses antibodies linked to radioisotopes to deliver targeted radiation to the bacteria and biofilm. We describe the first in vivo use of targeted radiation therapy, employing Actinium-225 (α-radiation) and Lutetium-177 (β-radiation) labeled antibodies to treat a Staphylococcus aureus biofilm-associated intramedullary implant infection. Untargeted radiation in the form of unbound radionuclide treatment was also evaluated. Methods: to assess therapeutic efficacy, bacterial counts were performed on implant and surrounding bone after seven days of follow-up. Biodistribution was evaluated using SPECT/CT and ex vivo gamma counting. Results: radioimmunotherapy using an antibody against wall teichoic acid which was labeled with Actinium-225 and Lutetium-177 achieved bacterial reductions between 45% and 93% on the implant and surrounding bone. Surprisingly, a similar antimicrobial effect was observed with unbound Actinium-225 treatment reducing the bacterial load by 80% on the implant and 98% in the surrounding bone. Indications of maximum tolerated dose (MTD) with Lutetium-177 labeled antibodies were observed through hepatic and renal function evaluations. Conclusions: These results should be interpreted in the context of the study’s constraints, particularly the limited animal sample size. Nonetheless, the results suggest that in vivo applied radiation may help reduce a biofilm-associated infection at the implant site as well as in the surrounding bone. These findings encourage further investigation into the use of targeted and non-targeted radiation, potentially combined with antibiotics, to develop effective strategies for eradicating biofilm-associated implant infections. Full article

The reliable and early detection of promising research directions is of great practical importance, especially in cases of limited resources. It enables researchers, funding experts, and science authorities to focus their efforts effectively. Although citation analysis has been commonly considered the primary tool to detect directions for a long time, it lacks responsiveness, as it requires time for citations to emerge. In this paper, we propose a conceptual framework that detects new research directions with a contextual Top2Vec model, collects and analyzes reviews for those directions via Transformer-based classifiers, ranks them, and generates short summaries for the highest-scoring ones with a BART model. Averaging review scores for a whole topic helps mitigate the review bias problem. Experiments on past ICLR open reviews show that the highly ranked directions detected are significantly better cited; additionally, in most cases, they exhibit better publication dynamics. Full article