Gustavus Adolphus College Collections
Not a member yet
92910 research outputs found
Sort by
Phrasing and prominence disambiguate clefted Relative Clauses
No full textWe investigated how prosody aids in disambiguating it-clefts with Connected Clauses We investigated the prosodic disambiguation of string-identical it-clefts with Connected Clauses (-Who sang? -It was [the editor] [that sang]) versus clefted Relative Clauses (-Who called? -It was [the editor [that sang]] ([that called])). Connected Clauses attach high in structure and convey background information, while clefted Relatives are nested within the focused element they modify and are also in focus. In the absence of prosodic cues, clefted Relatives tend to induce garden-path effects due to a default parsing preference for Connected Clauses. A production study revealed distinct prosodic phrasing and prominence patterns between the two structures across multiple regions, with disambiguation detectable from the first ambiguous word. Connected Clauses were prosodically separated from the clefted element, whereas clefted Relatives formed a single prosodic phrase with the head noun. These patterns align with their syntax, suggesting that syntax–prosody mismatches involving Relative Clauses are more constrained than previously assumed. In terms of prominence, clefted Relatives showed localized focus effects on the rightmost stressable word, rather than across the entire focused phrase. This supports the existence of intermediate representations linking information structure and prosody. An auditory comprehension study showed that listeners used these prosodic cues to override the default parsing preference for Connected Clauses
Relationships between heritable dementia risk factors, cardiovascular risk factors in young adulthood, and midlife neuropsychological outcomes
No full textBackground
Selected cardiovascular factors, APOE4 carriership, and family history (FH) are robust risk factors for Alzheimer’s disease and dementia. While cardiovascular risk tends to affect cognition from midlife, it remains unclear whether heritable risk predicts cardiovascular health in young adulthood and midlife, and whether young-adult cardiovascular health predicts midlife cognition.
Objective
We sought to examine how heritable dementia risk relates to cardiovascular health and how these cardiovascular risk factors in young adulthood predict midlife brain volumes and cognition.
Methods
We used data from the CARDIA study, which followed 5115 individuals aged 18-30 at baseline over 30 years. Analyses focused on 2808 participants (Mean age = 60, SD = 3.58) who attended the 30-year visit. We examined associations between APOE4 and FH with baseline and 30-year follow-up measures of cardiovascular risk factors (LDL-C, HDL-C, glucose, blood pressure, body mass index (BMI), smoking), cognition, and brain volumes.
Results
APOE4 carriers with FH had higher LDL-C and lower HDL-C levels as early as young adulthood, persisting into midlife. BMI and smoking were the only cardiovascular risk factors from young adulthood that predicted midlife cognition. There was no association between young adult cardiovascular risk factors and midlife brain volumes, but those with heritable dementia risk had larger brain volumes in regions vulnerable to midlife atrophy.
Conclusions
APOE4 carriership was associated with an unfavourable lipid profile that started in early adulthood and persisted to later life. Early cardiovascular risk was also associated with midlife cognition, which is earlier than studies typically focusing on later-life cognition
Nested resolution mesh-graph CNN for automated extraction of liver surface anatomical landmarks
No full textThe anatomical landmarks on the liver (mesh) surface, including the falciform ligament and liver ridge, are composed of triangular meshes of varying shapes, sizes, and positions, making them highly complex. Extracting and segmenting these landmarks is critical for augmented reality-based intraoperative navigation and monitoring. The key to this task lies in comprehensively understanding the overall geometric shape and local topological information of the liver mesh. However, due to the liver’s variations in shape and appearance, coupled with limited data, deep learning methods often struggle with automatic liver landmark segmentation. To address this, we propose a two-stage automatic framework combining mesh-CNN and graph-CNN. In the first stage, dynamic graph convolution (DGCNN) is employed on low-resolution meshes to achieve rapid global understanding, generating initial landmark proposals at two levels, “dilation” and “erosion”, and mapping them onto the original high-resolution surface. Subsequently, a refinement network based on mesh convolution fuses these landmark proposals from edge features along the local topology of the high-resolution mesh surface, producing refined segmentation results. Additionally, we incorporate an anatomy-aware Dice loss to address resolution imbalance and better handle sparse anatomical regions. Extensive experiments on two liver datasets, both in-distribution and out-of-distribution, demonstrate that our method accurately processes liver meshes of different resolutions, outperforming state-of-the-art methods. The reconstructed liver mesh dataset and the source code are available at https://github.com/xukun-zhang/MeshGraphCNN
Friction and wear reduction by glycerol oleates: The molecular basis for performance variations in the presence of water and acetic acid
No full textFriction and wear reduction by pure and technical grade glycerol monooleates (GMOs) was investigated in commercial and model base oils, including the influence of water and acetic acid (AA) impurities. Small-angle X-ray scattering (SAXS) indicated they cause micelle swelling and elongation. AA reduces the separation of OFM from oil at room temperature (RT). Tribotests in a mini-traction machine with white light interferometry (WLI) show that OFMs decrease the traction coefficient, especially at higher temperatures. AA addition also lowered the traction coefficient at RT. X-ray photoelectron spectroscopy (XPS) revealed surface chemical changes that depend on both tribotest conditions and lubricant composition. Wear was associated with the oxidation of metallic iron. The effects of impurities intrinsic to technical GMO appeared to be significant
Leveraging large language models for thematic analysis: a case study in the charity sector
No full textThis study explores how large language models (LLMs) can support deductive and inductive thematic coding in real-life contexts, balancing AI-driven efficiency with essential human oversight. Using three datasets from Tearfund, a UK-based Christian charity, we propose a dual-role human–LLM collaborative framework where the LLM functions as an initial annotator and a validator. In the deductive phase, GPT-4o and GPT-4o-mini were compared against human coders. GPT-4o achieved a substantial agreement in multi-label thematic categorization (κ = 0.61–0.65), while GPT-4o-mini showed a moderate agreement (κ = 0.41–0.58). Both models excelled in sentiment analysis (κ = 0.91–0.95), but struggled with evaluating evidence of impact due to contextual complexity (κ ≤ 0.01). GPT-4o-mini exhibited greater output variability and instability than GPT-4o, but benefited more from few-shot learning to mitigate hallucinations. In the inductive phase, GPT-4o demonstrated a strong semantic alignment with human-generated themes (cosine similarity = 0.76–0.79) though its tendency toward broad themes required human refinement. Despite their potential to streamline thematic analysis, LLMs also pose limitations and implementation challenges, including inconsistencies in excerpt extraction (precision = 0.41, recall = 0.53) and the trade-off between the time saved in coding and the time required for human validation. To facilitate practical implementation, we provide reusable prompt templates for four stages: context, instructions, data processing, and verification. Our findings underline the indispensable role of human expertise—from prompt engineering and managing hallucinations to final verification—to ensure accurate and trustworthy AI-assisted analyses. While LLMs can enhance qualitative analysis, their full potential is only realized under skilled human guidance
Dual-mode E-band mixer for 6G mmWave front-ends: simple structure with reconfigurable PT/resistive operation for optimized gain and linearity
No full textA dual-mode E-band mixer with a novel yet simple FET-based structure is designed and simulated using a 100 nm pHEMT PH10 process. The mixer employs a switchable impedance network and reconfigurable DC biasing to enable two distinct operational modes: a pump transconductance (PT) mode for higher conversion gain and a resistive mode for enhanced linearity. The results show that the mixer achieves a conversion gain of -4.2 dB in PT mode and -15.9 dB in resistive mode. In PT mode, with a 5 dBm LO drive, the input third-order intercept point (IIP3) is 1 dBm, and the input 1-dB compression point (IP1dB) is -4 dBm. In resistive mode, with a 2.6 dBm LO drive, the mixer achieves an IIP3 of 6 dBm and an IP1dB of 4 dBm. These results highlight the potential of the proposed design for 6G mmWave front-ends that require dynamic performance optimisation, such as balancing the trade-off between gain and linearity
Relating tribology to astringency perception in acidic plant protein-fortified fiber-based smoothies
No full textWith increased need to address environmental sustainability, there has been a pronounced interest on incorporating plant proteins in health-promoting fiber-rich fruit based drinks. Often such matrices are acidic in nature posing challenges for incorporating plant proteins causing undesirable textural issues such as astringency, which is poorly understood in the literature. This study aimed to understand how tribological and rheological characterization can help to explain mouthfeel of plant proteins when incorporated in fiber-based matrices (both model and real smoothies) at pH 3.8. Ten different commercially available isolated plant proteins (5 wt% protein solutions) exhibited significant aggregation being close to their isoelectric point in the fiber-based model smoothie dispersion (0.3 wt% pectin, 0.8 wt% inulin). Particularly, the viscosity of model smoothies spanned across three orders of magnitude, with many, if not, most demonstrating shear-thinning behaviors. Plant proteins exhibited diverse frictional dissipation, with some of the tested commercial fava bean protein, pea protein and chickpea protein concentrates outperforming industry standards, such as soy protein isolate. Model smoothie’s effectively mimicked real smoothies in mouthfeel attributes (11 trained panelists), showing plant proteins governing the mouthfeel. Pearson’s correlation identified strong relationships between boundary friction, rheology, and sensory attributes, highlighting the predictive value of in vitro methods. Notably, %soluble fraction negatively correlated with all tested undesirable attributes, such as astringency offering a facile screening metric for plant protein performance. Overall, this study validates the use of in vitro tools for mouthfeel assessment in complex food matrices, streamlining protein selection for accelerating the development of sustainable plant-based foods
Where tests fall short: empirically analyzing oracle gaps in covered code
No full textBackground: Developers often rely on statement coverage to assess test suite quality. However, statement coverage alone may only lead to 10 % fault detection, necessitating more rigorous approaches. While mutation testing is effective, its execution and human analysis costs remain high. Identifying covered statements that are not checked by oracles (e.g., assertions) offers a cost-effective alternative; however, the lack of empirical evidence for selecting the appropriate Oracle Gap Calculation Approach (OGCA) prevents developers from making informed choices. Aims: This knowledge-seeking study compares oracle gap characteristics determined by different OGCAs to assist developers in choosing the most valuable approach for their use cases. Method: Using mixed-method empirical analysis, we conduct an in-depth evaluation of the oracle gaps produced using three OGCAs: Checked Coverage using a Dynamic Slicer (CCDS), Checked Coverage using an Observational Slicer (CCOS), and Pseudo-Tested Statement Identification (PTSI). Across 30 Java classes from six open-source projects, we report on a quantitative evaluation of gap prominence, distribution, fault detection correlation and execution times, as well as results from a qualitative manual inspection of the statement types found in the oracle gaps. Results: The qualitative analysis showed data-loading statements, iteration statements and output updates to be most prominent in the oracle gaps. PTSI identified the oracle gaps with the lowest median mutation score (0.32), highlighting areas requiring more fault detection improvement compared to CCDS(0.76) and CCOS(0.50). PTSI also had the shortest median execution time (19.9 seconds), far quicker than both CCDS (273.2 seconds) and CCOS (5957.1 seconds). Conclusions: PTSI quickly reveals the priority testing areas for improved fault detection, making it an effective OGCA for developers to identify where tests fall short
