Gustavus Adolphus College Collections
Not a member yet
92910 research outputs found
Sort by
Alignment between intended and enacted pedagogies: a study of ELT curriculum innovation implementation in Pakistan.
No full textThis study investigates the alignment between English language education (ELE) pedagogy in policy (the pedagogical practices outlined in the ELE national curriculum) and pedagogy in practice (the pedagogical methods implemented by teachers in the classroom) at the secondary level (grades 9–10) in public schools in Punjab, Pakistan. The study is contextualised within the framework of ELE reforms, which were vital components of the broader Education Sector Reforms programme, initiated in Pakistan between 2001 and 2005. As part of these ELE reforms, a revised curriculum for English language instruction was introduced, promoting a comprehensive set of pedagogical principles that prioritise communicative, learner-centred, and inductive teaching approaches. Thirty-six English language lessons by twelve teachers were observed to assess their adherence to the pedagogical practices stipulated by the national curriculum. Additionally, post-observation interviews were conducted with the teachers to explore their reasoning behind the pedagogical strategies they employed or avoided in their instruction. The findings reveal a low level of compliance (29 %) with the recommended pedagogical policy. Key factors contributing to this compliance gap include exam-related pressures, institutional challenges, infrastructure limitations, and students’ low proficiency in English. The study has important implications for education policymakers, curriculum developers, administrators, and teachers
Spatio-temporal data fusion framework based on large language model for enhanced prediction of electric vehicle charging demand in smart grid management
No full textAccurate prediction of electric vehicle (EV) charging demand is pivotal for effective smart grid management and renewable energy integration. However, predicting spatio-temporal EV charging patterns remains challenging due to complex data fusion requirements arising from heterogeneous temporal, spatial, and contextual features, as well as difficulties in effectively integrating multiple modeling approaches. This paper introduces EV-STLLM, a novel spatio-temporal data fusion framework based on Large Language Model explicitly designed for accurate short-term EV charging demand forecasting through innovative integration of data-level and model-level fusion techniques. At the data level, a multi-source embedding module is developed to seamlessly fuse temporal features (e.g., time slots, weekdays), spatial heterogeneity (e.g., geographical location), and contextual charging behaviors into a unified representation via embedding convolutional network. At the model level, a large language model (LLM) is employed to capture global spatiotemporal dependencies, enhanced with Low-Rank Adaptation (LoRA) for parameter-efficient fine-tuning, substantially reducing computational costs while maintaining prediction robustness. Using a comprehensive real-world dataset comprising over 830,000 EV charging records across 16 districts and 331 subdistricts in Beijing, we validate EV-STLLM across multiple forecasting scenarios (district and subdistrict levels, one-step and two-step ahead predictions). Extensive comparative evaluations demonstrate that EV-STLLM consistently outperforms classical, graph-based, and deep learning baselines. Specifically, in one-step ahead district-level forecasting, EV-STLLM achieves up to a 15.41% reduction in MAE and a 53.51% reduction in MAPE compared to the leading baseline, underscoring its potential to significantly enhance data-driven smart grid operations
Influence of sleeper geometry on the lateral resistance of bamboo sleepers
No full textThis study investigates the effects of bamboo sleeper geometry on lateral track resistance in ballasted railway systems using the Discrete Element Method (DEM), where ballast is modeled as clumped spherical particles with Hertz-Mindlin contact model for non-linear stiffness. Four sleeper geometries, including traditional rectangular, dumbbell, rectangular-winged, and wedge-winged, were analyzed to evaluate their interaction with ballast under lateral loading. The Single Tie Push Test (STPT) procedure was used to measure lateral resistance, with lateral forces applied via a hydraulic jack. The results highlight that sleeper geometry significantly affects lateral resistance, with optimized designs (rectangular- and wedge-winged) achieving the highest resistance values. For the rectangular-winged sleeper, lateral resistance peaked at 6.85 kN, representing a 54% improvement over the traditional rectangular shape. The study also examines the contributions of ballast components (base, crib, and shoulder) to the overall lateral resistance. Base resistance dominated for all geometries but decreases as the crib and shoulder contributions increase with the non-traditional designs. Normal and tangential force distributions within the ballast were also analyzed, showing enhanced interparticle contact for the winged designs. Overall, the rectangular-winged sleeper was found to provide higher performance compared to the traditional prototype sleeper with respect to meeting track requirements for lateral stability
Improving collaborative engagement in health state valuation: a scoping review of current practices and emerging recommendations
No full textBackground and Objective
Collaborative engagement with individuals invested in or affected by health research, beyond researchers themselves, is advantageous and encouraged by major funding bodies. However, the degree of collaborative engagement in health state valuation is unclear. A scoping review was conducted to (i) identify recommendations on best practice in collaborative engagement in health economics and related literature; (ii) identify examples of collaborative engagement in valuation studies; and (iii) map (ii) onto (i) to identify current practice and future recommendations.
Methods
Eight databases were searched in March-May 2024, with grey literature searches in August-September 2024. For objective (i), reports or manuscripts in health economics or patient-reported outcome measure development/evaluation of any date providing recommendations for collaborative engagement were included. For objective (ii), articles published since 2019 featuring health state valuation and collaborative engagement were included. Best practice recommendations were extracted and thematically synthesised. Examples of collaborative engagement were extracted and mapped against recommendations.
Results
Twenty-two records featuring recommendations and 15 valuation studies were included. A 15-item framework of emerging best practice recommendations for collaborative engagement was synthesised. Most examples of collaborative engagement involved patients and/or experts helping inform health states for valuation. There was no evidence for 9 out of 15 synthesised recommendations having been applied in any of the valuation studies and only minimal evidence was extracted for the remaining six.
Conclusions
Collaborative engagement in health state valuation is underdeveloped and unaligned with literature recommendations. A 15-point framework has been developed as a strategic starting point for developing guidance to improve practice in the field
Organic geochemical investigations of an MIS 5 fire in the Palaeolithic deposits of Ormesson (Seine-et-Marne, France):Anthropic or natural?
No full textDespite the central role of fire in Pleistocene and Palaeolithic lifeways, the relationship among hominins, fire, and their environment remains unclear. Ancient combustion residues hold a wealth of molecular data that may help to resolve some of these questions, yet standardised guidelines for reconstructing past fire traces are notably lacking. In this study, we examine extensive combustion residues overlying Middle Palaeolithic deposits from the open-air site of Ormesson (France). To determine whether the combustion residues are of natural or human origin, multiproxy approaches including anthracology, lipid biomarker, and benzene polycarboxylic acid (BPCA) analyses are applied. These techniques are used to characterise organic matter and pyrogenic carbon compositions in the deposits, providing insights into surrounding vegetation, palaeoenvironmental shifts, and the production parameters involved in the formation of the char assemblage. Lipid biomarker evidence suggests that the pre-fire local environment featured abundant coniferous vegetation (e.g., Pinaceae taxa), which is supported by anthracological evidence for a predominance of Pinus cf. sylvestris/nigra complemented by Betula sp. taxa. The post-fire environment saw a contraction of coniferous vegetation, concurrent with an expansion of deciduous taxa, grasses and herbaceous material. The combustion event, which resulted in 67 % of the charcoal assemblage exhibiting vitrification, produced PyC contents of up to 362 g/kg OC in soil samples and 443 g/kg OC in charcoal samples, with aromatic condensation values of up to 34 %. BPCA-derived predictions of heat treatment temperatures yielded values of approximately 300–400 °C for charcoal samples and 400–550 °C for soil samples in the burned layer, constituting the first instance in which quantitative temperature estimations are obtained from BPCA results. Based on the integrated evidence, we accept the null hypothesis that the studied combustion residues are natural in origin. However, the similarity of archaeometric and geochemical signatures from natural and human-controlled fires underscores the need for interdisciplinary, multiproxy efforts to improve the identification of past fire regimes
Time-resolved SAXS studies during the synthesis of hydrolytically degradable poly(ε-caprolactone)-poly(N,N′-dimethylacrylamide) diblock copolymer nanoparticles in aqueous media
No full textThe mechanism of nanoparticle formation during reverse sequence polymerization-induced self-assembly (PISA) is studied by small-angle X-ray scattering (SAXS). More specifically, N,N′-dimethylacrylamide (DMAC) monomer is added to a trithiocarbonate-capped poly(ɛ-caprolactone) (PCL) precursor and initially polymerized in the bulk at 80 °C via reversible addition-fragmentation chain transfer (RAFT) polymerization. SAXS indicates the unexpected formation of molten PCL droplets dispersed within DMAC monomer. After 5 min (14% DMAC conversion) at 80 °C, the reaction mixture is diluted with water, and the aqueous milieu is analyzed using a flow cell. A transient lamellar phase is formed immediately after water addition that subsequently transforms into nascent spherical nanoparticles. During the remaining DMAC polymerization, the overall nanoparticle diameter remains essentially constant with a concomitant reduction in the PCL core radius and the aggregation number. This suggests that individual PCL-PDMAC chains are in equilibrium with the nanoparticles. SAXS analysis indicates that the amorphous PCL cores have a mean diameter of 8.8 nm at 80 °C: X-ray diffraction (XRD) studies confirm that such nanoscale confinement prevents their crystallization on cooling to 20 °C. Finally, this formulation can be combined with crystallization-driven self-assembly (CDSA): UV-initiated DMAC polymerization at 15 °C produces rod-like PCL-PDMAC nanoparticles with semicrystalline PCL cores
Machine learning in peak demand forecasting: foundations, trends, and insights
No full textPeak demand forecasting involves predicting the maximum electricity demand within a specific period, which plays a key role in maintaining the efficiency and stability of power systems. The rapid evolution of power systems, driven by advanced metering infrastructure, local energy applications such as electric vehicles, and the increasing adoption of intermittent renewable energy, has introduced greater randomness and reduced predictability in peak demand. Given the pressing need to address more diverse implementation requirements across different contexts, accurate and reliable peak demand forecasting has become increasingly important. To the best of our knowledge, this study is the first to provide a comprehensive overview of peak demand forecasting methods. It systematically reviews 186 studies published since the 1950s, categorizing these methods into three stages based on their developmental timeline. Building on this, the study defines a unified framework for peak demand forecasting and offers an in-depth analysis linking these methods to the practical needs of power systems. Notably, it highlights the growing importance of machine learning-driven forecasting models in addressing the increasing complexity of modern energy environments. Furthermore, this study identifies key research gaps and points out emerging trends that hold potential for advancing innovation in this field
Task-Free Continual Generative Modelling Via Dynamic Teacher-Student Framework
No full textContinually learning and acquiring new concepts from a dynamically changing environment is an important requirement for an artificial intelligence system. However, most existing deep learning methods fail to achieve this goal and suffer from significant performance degeneration under continual learning. We propose a new unsupervised continual learning framework combining Long- and Short-Term Memory management used for training deep learning generative models. The former memory system uses a dynamic expansion model (Teacher), while the latter uses a fixed-capacity memory buffer to store the update-to-date information. A novel Teacher model expansion approach, called the Knowledge Incremental Assimilation Mechanism (KIAM) is proposed. KIAM evaluates the probabilistic distance between the already accumulated information and that contained in the Short Term Memory (STM). The proposed KIAM adaptively expands the Teacher's capacity and promotes knowledge diversity among the Teacher's experts. As Teacher experts, we consider generative deep learning models such as~: the Variational Autocencoder (VAE), the Generative Adversarial Network (GAN) or the Denoising Diffusion Probabilistic Model (DDPM). We also extend the KIAM-based model to a Teacher-Student framework in which we use a data-free Knowledge Distillation (KD) process to train a VAE-based Student without using any task information. The results on Task Free Continual Learning (TFCL) benchmarks show that the proposed approach outperforms other models
A fully automated explainable predictive model for diagnosing pre-capillary and post-capillary pulmonary hypertension on routine unenhanced CT: results from the ASPIRE registry
No full textAims
Unenhanced chest CT is frequently used to assess lung malignancy and parenchymal disease. Harnessing CT data to quantify cardiac and vascular structures has the potential to improve the diagnosis of heart failure and pulmonary hypertension (PH). This study aims to develop a deep learning model to segment and analyse cardiothoracic structures from unenhanced CT images to diagnose PH, pre-capillary PH and PH associated with left heart disease (LHD).
Methods and results
A twelve-structure cardiothoracic segmentation model was developed using an institutional cohort (n = 55, 35/9/11 training/validation/testing). Model performance was evaluated using Dice similarity coefficients (DSC). Volumetric measurements were compared to manual values using intra-class correlation (ICC) and visually assessed by four observers using an external cohort (n = 50, from 26 hospitals). Univariable and multivariable regression analyses were performed using a cohort of 368 patients (254/114 training/testing). Receiver-operating characteristic curves were plotted and the area under the curves (AUC) with confidence intervals (CI) were calculated. The model yielded a DSC segmentation performance of ≥0.87 for 9/12 segmented structures and ICC > 0.95 for 10/12 structures. Most of the segmented structures scored as excellent in the external cohort visual assessment. Diagnostic accuracy for predicting PH was high [AUC = 0.88 (CI: 0.80–0.96), sensitivity = 70%, specificity = 100%], including pre-capillary PH [AUC = 0.84 (CI: 0.74–0.94), sensitivity = 72%, specificity = 94%] and PH-LHD [AUC = 0.86 (CI: 0.79–0.93), sensitivity = 94%, specificity = 63%].
Conclusion
A fully automated model for multi-structure cardiothoracic segmentation on unenhanced CT is achievable. The model can predict PH and identify patients with pre-capillary PH and PH-LHD with promising performance
A structured ultrasound‐guided cannulation course to prepare medical students for foundation training
No full textBackground
Ultrasound-guided peripheral intravenous cannulation (US-PIVC) is a critical skill for resident doctors, yet standardised ultrasound training remains inconsistent in undergraduate medical curricula. This gap may compromise patient care and safety.
Approach
A structured, competency-based US-PIVC simulation training was integrated into the final-year medical curriculum. Using the Reach, Effectiveness, Adoption, Implementation and Maintenance (RE-AIM) framework, we conducted a convergent parallel mixed-method study. Quantitative data were collected through a validated rating scale in an end-of-session assessment, whereas qualitative insights were gathered via focus group discussions.
Evaluation
Ninety-eight students participated in the simulation training, with students (n = 25) and staff (n = 4) contributing to focus group discussions. The objective competency assessment demonstrated a 98% pass rate, with 84% achieving full procedural proficiency. Thematic analysis revealed that structured US-PIVC training significantly enhanced students' confidence and preparedness for their foundation doctor role. Participants reported a perceived reduction in dependence on senior staff and improvements in both patient safety and procedural efficacy. To ensure skill retention, key recommendations included providing ongoing practice opportunities, implementing logbook signoffs, appointing designated US skills leads, fostering collaborative partnerships and maintaining US equipment.
Implications
Our study highlights the need for structured, standardised US-PIVC training to reduce variability in clinical education. The programme improved confidence, proficiency and clinical efficiency while decreasing reliance on senior staff. Embedding mandatory training, logbook signoffs and simulation realism will enhance patient safety, procedural competency and preparedness for foundation roles