26734 research outputs found
Sort by
Forecasting futures for a coastal broad-leaf evergreen tree in decline, Arbutus menziesii: Comparing subpopulations to whole species responses for climate change and Pacific Madrone Leaf Blight risk.
No full textArbutus menziesii is a coastal evergreen tree observed to be in decline over the last 50 years. The cause of its decline has been associated with climate change resulting in a reduction in its habitat and the increased presence of fungal pathogens, yet the relative importance, and temporal effects of these threats are unknown, and considered variable throughout the range of this species. By examining subpopulations across A. menziesii’s range for forecasted suitable habitat and Pacific Madrone Leaf Blight (PMLB) risk, I assessed the relative range-wide, temporal effects and potential contributions to decline. Subpopulation delineation is a challenge for species lacking range-wide genetic information, like A. menziesii. This thesis used a novel subpopulation clustering method integrating both isolation by distance and environment to establish coarse-scale subpopulations. Data for subpopulation clustering and forecasting were sourced from the Global Biodiversity Information Facility (GBIF) (n = ~ 18 000) offering range-wide representation of A. menziesii. Herbarium data also sourced from the GBIF (n = ~600) spanned the last 120 years, also offering range-wide representation of A. menziesii. Four subpopulation delineation methods were compared using geographic and environmental isolation clusters. The highest performing subpopulation model used both isolation by distance and isolation by environment and could better discriminate presence and background points at the local scale compared to the whole species model and other subpopulation delineation method models. The PMLB scored herbaria dataset was combined with ClimateNA climate variables from the historical and 13 GCM ensembled SSP3-7.0 future time series to model blight risk from 1901-2100 for whole species and subpopulations. Chelsa historical and future climate bioclimatic predictors from SSP3-7.0 were used to build an SDM and forecast suitable habitat for whole species and subpopulation models. Whole species and subpopulations differed in their blight risk and forecasted suitable habitat. Subpopulation models produced a more optimistic suitable habitat forecast, with a percent decrease compared to historical of 3%. Whole species models produced less optimistic suitable habitat forecasts, with a percent decrease compared to historical of 12%. Both models show southern range contraction and northern range expansion, though the subpopulation model predicts more north-central area. Whole species blight risk from historical herbaria data shows little change in blight prevalence from 1901-2023, however subpopulations had significantly different blight prevalence. Northern subpopulations had the highest blight prevalence and southern subpopulations had the lowest blight prevalence. Whole species forecasting found blight risk unchanged but was significantly different for subpopulations through time. Blight risk is predicted to decrease for northern subpopulations, increase for southern subpopulations and stay unchanged for central subpopulations. Comparing both the forecasted species distribution models and PMLB risk maps suggests that southern populations are likely to be under double the pressure of both climate change moving them out of their preferred habitat, and increased PMLB presence where they historically experienced a lower incidence. Given the predicted shift north of suitable habitat for species, safeguarding the southern population for translocation may be important if southern conditions transfer northward to reduce the decline of A. menziesii
Comparisons between Clinician and AI Predictions of Patient Outcomes in the Intensive Care Unit
No full textAccurate prediction of patient outcomes in the intensive care unit (ICU) is essential for improving patient care, optimizing resource allocation, and supporting clinical decision-making. Although machine learning (ML)-based artificial intelligence (AI) models have demonstrated strong predictive performance in retrospective studies, direct comparisons with human clinicians, particularly in prospective, real-world settings, remain limited. This study addressed this gap by comparing the predictive performance of ML models and clinicians for multiple ICU outcomes in both retrospective and prospective settings. Retrospective clinician predictions were obtained for 990 ICU admissions from Alberta’s 15 adult tertiary ICUs, and prospective clinician predictions were collected at the bedside for 238 ICU admissions between 16 and 24 hours after ICU admission. ML models were trained on data from 46,631 ICU admissions (41,096 unique patients) and evaluated on the same cases assessed by clinicians. Predicted outcomes included in-hospital mortality, 30-day post-discharge mortality, ICU and hospital length of stay, delirium, and acute kidney injury. ML models generally outperformed individual clinicians across outcomes in the retrospective setting, although clinician performance sometimes exceeded ML when predictions were aggregated. In the prospective setting, subspecialized physicians tended to match or exceed ML performance, whereas trainees and nurses generally performed worse than AI. Prediction of the length of stay outcomes were poor for both clinicians and ML, and inter-rater agreement was consistently low to fair both among clinicians and between clinicians and ML. These findings provide a comprehensive benchmark for clinician and AI performance in ICU prognostication and clarify the distinct strengths of human and ML-based prediction. Together, these findings underscore the need for prospective evaluation when assessing clinical AI tools and highlight the complementary roles that clinician expertise and data-driven models may play in critical care
On the Kinetic Elucidation of the Synthetic Mechanisms of Silver Nanoparticles (AgNPs) for Application in Localized Surface Plasmon Resonance (LSPR) Based Enhancement of Triplet State Quantum Yields
No full textThe use of plasmonic silver nanoparticles (AgNPs) in relation to catalysis, medicine, and new emerging solid-state technologies has been widely explored in the past two decades and is a culmination of a century of discovery and application. However, the replicability of AgNPs syntheses is inconsistent. It is reagent batch dependent at best and irreproducible at worst. This thesis explores a streamlined method of synthesizing AgNPs by adapting the Tollen’s reagent at optimized concentrations to obtain a facile, robust, and reproducible, room-temperature seed-mediated synthesis with a high degree of size control. The role of solvent mixtures in this novel synthesis was investigated with a conclusion that acetonitrile plays a role in inhibiting nucleation when compared to pure water. Though the role is not fully elucidated, the acetonitrile may allow for more uniform nanoparticle growth at higher concentrations of seeds as the kinetics of seed growth do not seem to change as the concentration of seeds increases. This optimized method of AgNPs synthesis was then tested with current functionalization protocols that include but are not limited to silica coating, amination, and conjugation of xanthene dyes. This was done to explore potential enhancements of photophysical processes, specifically with respect to the enhancement of singlet oxygen production. This part of the thesis is inconclusive. However, an optimization in the testing of dye conjugation is proposed with potential time savings for future work
Functional Comparison of Synovial Membrane MSCs and iPSC-Derived iMSCs from Osteoarthritis-Diagnosed Patients
No full textOsteoarthritis (OA) is a chronic disease characterized by progressive cartilage degeneration, subchondral bone changes, and synovitis, leading to joint dysfunction. Although stem cell-based therapies offer promise, synovial membrane-derived mesenchymal stem cells (SM-MSCs) from OA patients display donor-dependent declines in proliferative and chondrogenic potential, restricting their utility as a cell source for regenerative medicine strategies. Induced pluripotent stem cells (iPSCs) provide a platform for scalable and patient-specific regenerative applications but concerns regarding genetic instability and tumorigenicity persist. Reprogramming endogenous SM-MSCs into iPSCs and subsequent derivation of induced MSCs (iMSCs) may mitigate age- and disease-related epigenetic drift, enabling the generation of individualized, clinically relevant cell products for OA treatment. This study systematically compares the regenerative and chondrogenic attributes of patient-matched SM-MSCs and iMSCs generated via two integration-free iPSC reprogramming methods: Sendai virus (SeV) transduction and episomal nucleofection using core pluripotency transcription factors. SM-MSCs were isolated from two OA-diagnosed patients (a female and male donor). Genomic integrity of derived iPSC lines was confirmed via G-banded karyotyping, while transgene clearance and pluripotency were verified through flow cytometry, immunocytochemistry, qualitative PCR, and in vivo teratoma assays. iMSC identity was validated by flow cytometry and in vitro trilineage differentiation potential. Both reprogramming strategies yielded iPSC colonies; however, SeV-derived lines, particularly from the male donor, exhibited compromised stability and increased spontaneous differentiation, restricting their progression to iMSC derivation. Only episomal iPSC lines maintained robust pluripotency and supported iMSC generation for both donors. Comparative analyses included in vitro chondrogenesis, global proteomic profiling, and functional analysis using an in vivo post-traumatic OA mouse model to assess cartilage repair efficacy of the cell populations. This study highlights the therapeutic potential of iMSCs while analyzing donor effects, reprogramming method selection, and the need for molecular validation in developing effective OA regenerative therapies. It contributes to our understanding of patient-specific cell sources and integration-free cellular reprogramming for translational joint repair
Dynamic Sparse Training Breaks the Correlation between Plasticity and Reward in Deep Reinforcement Learning
No full textTraining deep reinforcement learning (DRL) agents remains brittle, a difficulty often attributed to a loss of network plasticity, commonly monitored using proxies such as dormant neuron prevalence and representational rank. Dynamic sparse training (DST), which continually prunes and regrows network connections, can be motivated as a mechanism for preserving plasticity under non-stationary learning problems. This thesis examines whether DST reliably improves standard plasticity proxies in value-based DRL, and whether such improvements align with cumulative reward. In a controlled supervised learning setting with induced non-stationary targets, we con"rm that DST effectively mitigates plasticity loss. Following this we conduct an empirical study comparing dense, static sparse, gradual magnitude pruning (GMP), and DST agents across modern value-based algorithms, including deep Q-network (DQN) and parallelised Q-network (PQN), on the Arcade Learning Environment (ALE). Our results reveal a consistent de-correlation between plasticity and reward. Although DST improves standard plasticity proxies most effectively, these improvements do not result in superior rewards. In contrast, GMP achieves the highest performance by maintaining intermediate plasticity levels, which surpass dense baselines but remain below those of DST. These results support the hypothesis that the mechanism of continual adaptation in DST can induce a level of plasticity that is detrimental to value-based learning. Ablations over sparsity and mask evolution hyperparameters show that topological changes with stronger plasticity proxies do not reliably increase return, the highest return regimes vary across environments, and that both overly conservative and overly aggressive topology updates can impair performance. These findings motivate future work on developing more effectively, environment dependent topological evolution strategies for DST in DRL
Real-Time Multiresolution Management of Spatiotemporal Earth Observation Data Using DGGS
No full textEarth observation (EO) data play a crucial role in environmental monitoring, climate analysis, and other large-scale geospatial applications. However, the rapid growth in the volume, spatial resolution, and temporal frequency of modern EO datasets has made their storage, processing, and access increasingly more challenging. Existing data management approaches struggle with scalability, geometric distortion, and redundancy. These challenges are further compounded by temporal gaps caused and missing data, and the lack of continuous temporal representations, as well as inefficient support for multiresolution querying and interactive access over arbitrary regions of interest. Together, these limitations hinder the practical use of EO data in time-sensitive and large-scale applications. To address these challenges, this thesis adopts a triangular Discrete Global Grid System (DGGS) as the foundation for a spatiotemporal data management framework. The hierarchical and globally consistent structure of DGGS enables efficient multiresolution representation and indexing of spatial data, while minimizing the geometric distortion that plagues traditional raster grids. Building on this foundation, the proposed framework introduces a tensor-based multiresolution storage scheme coupled with a triangular wavelet scheme defined directly on the DGGS grid. To handle temporal gaps, continuous temporal approximation is achieved using reverse Chaikin subdivision and B-spline curve fitting, allowing the system to not only support smooth approximation and live updates, but also to significantly reduce storage requirements. Together, these components define a scalable framework for real-time multiresolution processing, storage, and retrieval of spatiotemporal raster data. The proposed framework is validated through a case study using data from the RADARSAT Constellation Mission, where experimental results demonstrate efficient encoding, dynamic retrieval, and interactive visualization of time-varying EO data for user-defined regions, while maintaining high data fidelity. Moreover, the framework's scalability and performance are further evaluated using two additional large-scale EO datasets; one spatially extensive dataset covering the entire Canadian region, and another temporally extensive dataset spanning nearly 7 years over Gull Lake, Alberta. The results highlight the framework's ability to handle large volumes of EO data while providing efficient storage and access capabilities
The automation of nanoHX for sub-pmol amounts of protein
No full textHydrogen–deuterium exchange mass spectrometry (HDX-MS) is a powerful technique for probing protein structure and dynamics; however, its broader application in pharmaceutical research has been constrained by limited sensitivity and high sample requirements. Our earlier nanospray HDX-MS system improved sensitivity but lacked automation, robustness, and online digestion. Here, I present a robust NanoHD source compatible with Thermo Scientific EasySpray mass spectrometers. The prototype enables a substantial reduction in LC separation flow rates to the nL/min regime, compared with conventional HDX-MS workflows operating at µL/min. This nanoflow LC configuration enhances analytical sensitivity. To support digestion strategies under these conditions, a miniaturized immobilized enzyme reactor (IMER) was developed for online digestion, and in-solution digestion was also evaluated to optimize digestion performance under NanoHD conditions. The results demonstrate that both IMER-based and in-solution digestion are compatible with nanoflow HDX-MS. The NanoHD source was subsequently integrated with an autosampler and the IMER, enabling fully automated operation with improved sensitivity, reduced sample consumption, and reliable routine use. Using this platform, we achieved near-complete sequence coverage with only 250 fmol of phosphorylase B and Eg5, and 0.5 pmol of membrane proteins such as dimeric BlaR1 and the TagH/TagG complex—approximately 100-fold less protein than required for conventional HDX-MS workflows. Low back-exchange (~25%) at these ultralow flow rates demonstrates the effectiveness of the system design and its suitability for routine and high-throughput applications. Fully automated measurements were further demonstrated by investigating the interaction between FEN1, a cancer-relevant target, and a small-molecule inhibitor. The NanoHD platform produced HDX-MS results for 0.5 pmol of FEN1 that were comparable to those obtained using approximately 80 pmol in a conventional workflow, representing a ~160-fold improvement in sensitivity. Overall, this automated nanoflow HDX-MS workflow substantially reduces sample requirements while maintaining high analytical performance, extending the applicability of HDX MS to scarce or complex proteins in biopharmaceutical and structural biology research
Expanding the implementation of virtual parent-led peer support groups for parents of children and adolescents with eating disorders: a convergent mixed methods study
No full textAbstract Background Parenting a child with an eating disorder (ED) while also navigating the healthcare system can be complex and overwhelming. Preliminary research indicates that parent-led peer support could be a promising avenue to ease this burden. This study examined the expanding implementation of virtual parent-led peer-support groups (vPLPSGs) for parents of children with EDs. Methods A convergent mixed methods design was used to evaluate the effectiveness of the vPLPSG intervention and success of its implementation. Parents of children who had recovered from an ED, or parent peer support providers (PPSPs), were trained to lead vPLPSGs for parents of children currently ill with an ED. Groups occurred biweekly across two study waves, totalling 12 months. Parents completed measures of caregiver burden, caregiver needs, and caregiver confidence before and after attending vPLPSGs over a three-month period. PPSPs completed weekly ratings of fidelity and pre- and post-implementation measures of readiness for change, attitudes towards evidence-based practice, and perceptions of the effectiveness of the principles guiding this intervention. Using qualitative methods, PPSPs and parents participated in a focus group and individual semi-structured interviews, respectively. Results At post-intervention, parents (n = 35) reported decreased caregiver burden (MD=-4.69; p = 0.050), an increase in met information and support needs (MD = 5.60; p < 0.001), and increased confidence to support their child with an ED (MD = 8.09; p < 0.01). Qualitative data indicated that parents reported vPLPSGs as a positive, supportive experience. PPSPs (n = 8) experienced a decrease in PPSPs’ attitudes towards evidence-based practices (MD=-0.54; p < 0.01). Around half (50.5%) of the fidelity ratings met the pre-determined criterion (80%). PPSPs gave overall positive qualitative reports of their experiences facilitating vPLPSGs. Conclusions In this expanded vPLPSG intervention implemented across Canada, the program demonstrated effectiveness in improving parent outcomes and supporting PPSPs’ readiness, attitudes, and perceived fit for the intervention. Both parents and PPSPs found the intervention acceptable and meaningful, and PPSPs demonstrated fidelity to the model. Parent-led peer support may be a useful adjunct to traditional pediatric ED care. Further research is needed to explore the implementation of vPLPSGs across diverse healthcare settings and demographic groups.Plain English summary Parents whose children have eating disorders face many challenges and need support. Support from other parents of children with eating disorders has been suggested as beneficial. One small study found good acceptability and effectiveness of virtual parent-led peer support groups for parents of children with eating disorders. The current study examined these groups on a larger national scale. Parents of children who had recovered from an eating disorder were trained as group leaders and facilitated groups across two study waves, totalling 12 months. Parents of children with a current eating disorder participated in the groups over a 3-month period. Caregiver burden improved as did confidence in supporting their child. Both the peer support group leaders and parents found the groups to be positive, supportive environments. Parent-led peer support shows promise as a part of the continuum of eating disorder care and merits further study on a broader scale
Association between sociodemographic factors and mobility among older adults: a systematic review and meta-analysis
No full textAbstract Background Mobility limitation is associated with poor quality of life, morbidity, and mortality among older adults. This pre-registered systematic review [PROSPERO CRD42022298570] synthesised the coefficients of association between sociodemographic factors and performance-based mobility outcomes in older adults (≥ 60 years). Methods Electronic databases MEDLINE, WoS, EMBASE, CINAHL, AgeLine, and SPORTDiscus were searched from inception to 27 November 2023 for observational studies reporting an association between sociodemographic factors and performance-based mobility outcomes among older adults. Pairs of reviewers independently conducted title, abstract, and full-text screening, narrative synthesis, and meta-analysis following the PRISMA and MOOSE guidelines. The effect sizes, heterogeneity, dominance, and publication bias were analysed using R/R-Studio (version 4.3.2) and CMA (version 4). Results Of the 9,328 studies screened, 57 were included (n = 130,060 participants); the pooled mean age was 69.81 ± 7.21years, habitual gait speed (HGS) = 1.01 ± 0.28 m/s, and time-up and go score = 7.67 ± 3.56s. The narrative synthesis showed that the majority of the studies found older age (92.2%), women (62.9%), non-Caucasian (75.0%), and lower education (64.5%) associated with significant mobility outcomes. There was a paucity of studies on marital status, area of residence, income, occupation, religion, homeownership, and social status. Meta-analysis showed that older age r=-0.37 [-0.42, -0.32] and female gender r=-0.13 [-0.22, -0.03] were moderately associated with slower HGS. Conclusion Older age, female gender, non-Caucasian identity, and lower education were consistently associated with poorer mobility outcomes, pointing to sociodemographic sources of inequity and the need for targeted interventions. Limited evidence on marital status, residence, income, occupation, religion, homeownership, and social status restricted broader conclusions. Expanding research across these domains is critical to inform comprehensive strategies that advance equitable mobility and healthier ageing in diverse populations. Trial registration Systematic review registration: PROSPERO CRD42022298570
Traffic Monitoring from Sensors to Systems: Optimizing Sensor Placement, Minimizing Estimation Errors, and Unifying Macroscopic Fundamental Diagrams
No full textTraffic monitoring is the backbone of traffic management and the essential ingredient for traffic control. With the advent of big data analytics and artificial intelligence applications in traffic, it seems prudent to revisit the components needed for a successful intelligent transportation system. This thesis takes concrete steps in this direction by focusing on the main steps of traffic monitoring (data acquisition, traffic state estimation, and macroscopic traffic estimation) and addressing some gaps in the literature. Namely, this thesis addresses the lack of a solid foundation for sensor location optimization considering macroscopic traffic monitoring, the lack of an effective real-time network-wide link-level model for traffic state estimation using sparse data, and the lack of a holistic understanding of macroscopic fundamental diagrams and what predicts their shapes. This dissertation addresses these gaps by developing a unified “sensors-to-systems” workflow to improve monitoring accuracy and interpretability. First, it formulates a network sensor location problem with an explicit objective of macroscopic monitoring accuracy and proposes a PCA-based approach that ranks links by newly defined criticality measures aided by clustering. The developed model is shown to significantly reduce errors compared to a benchmark. Second, it introduces a real-time state estimation framework that tracks waves through merges and diverges and retroactively corrects estimates. Both a toy network and a large simulated network from Philadelphia verify the accuracy of the model. Third, the dissertation investigates whether MFD shape can be predicted from non-traffic features and whether a near-universal MFD form can be derived. A GNN is trained to predict the uncongested branch of the MFD, but results show strong asymmetry between flow and density predictability, indicating data and model depth limitations and a fundamentally harder density-prediction problem. Complementary statistical analysis finds that flow-related targets are relatively explainable while density at capacity is only moderately predictable. Feature importance measures highlight socioeconomic variables, urbanization proxies, environmental conditions, and detector placement as key predictors. Finally, based on these insights and parallels from ecological research, an interpretable hybrid analytical MFD is proposed that blends cubic and exponential behavior using an urbanization-dependent transition. It outperforms classical link-level FD forms across diverse networks