Faculty of Graduate Studies Online Theses

Permanent URI for this collectionhttps://hdl.handle.net/10222/11163

Browse

Now showing 1 - 20 of 8515

Access status: Open Access ,
The Prevalence of Substance Use in Injured Off-Road Vehicle Drivers: A Systematic Review and Observational Study
(2025-12-15) Rheault, Grace; Not Applicable; Master of Science; Department of Community Health & Epidemiology; Received; NA; Not Applicable; Dr. Jill Hayden; Dr. Mark Asbridge; Dr. David Clarke
Background: Drug driving has become as common as alcohol-impaired driving in Canada and is a major public health issue. Substance use in the context of off-road vehicle driving (i.e., ATVs, snowmobiles, dirt bikes) has not been well-studied in Canada. The prevalence of acute substance use in injured off-road vehicle drivers across Canada is currently unknown. The aim of this study is to estimate the prevalence of acute substance use among injured off-road vehicle drivers in Canada. Objectives: This thesis includes two main studies and three key objectives. Part 1: Systematic Review and meta-analysis The first study involves a systematic review and meta-analysis, with the purpose of (1) estimating the prevalence of acute substance use in drivers involved in off-road vehicle crashes. Part 2: Observational Study The second study is a primary study of injured off-road vehicle drivers presenting to emergency departments across Canada. The two objectives of this study are: (2) to estimate the prevalence of acute substance use in injured off-road vehicle drivers presenting to emergency departments in Canada; and (3) to determine what factors (i.e. demographic, crash characteristics, regional variation) are associated with acute substance use in injured off road vehicle drivers presenting to emergency departments in Canada. Systematic Review Methods: Studies were identified from electronic databases and grey literature. Study selection, data extraction, and risk of bias assessment were completed by two independent reviewers. Descriptive statistics were used to report study characteristics. Random effects meta-analyses were used to report the overall pooled prevalence, and subgroup and sensitivity analyses were used to explore study characteristic effects and the robustness of results. Primary Study Methods: Employing data from the National Drug Driving Study, this study analyzed data from injured drivers presenting to 18 participating EDs across Canada who were in an off-road vehicle crash and had blood drawn for clinical purposes. Excess blood from clinical use underwent toxicology analysis to quantify the presence of impairing substances in the injured driver. Prevalence estimates of substance use in injured off-road vehicle drivers were reported overall, by substance type, by level of alcohol and tetrahydrocannabinol (THC), by number of substances and disaggregated by relevant covariates. Results: 20 publications comprising of 18 studies were included for review, with 2 sets of publications that had overlapping data, drawn from the same studies. Prevalence estimates ranged from 8% to 85%. The overall pooled prevalence estimate of substance use was 41% (95% CI: 28%-54%). There was significant heterogeneity across studies (Q = 3456.38, p < 0.001; I2=99.4%). Subgroup analyses found significant differences between studies using different data sources (coroner’s data: 57% (43%-71%) vs. hospital data (29% (6%-58%), and between non-fatally (29%, 95% CI: 9%-55%) and fatally injured drivers (56%, 95% CI: 43%-69%). The observational study included 473 injured drivers who met the inclusion criteria. Overall, 71% of tested drivers were positive for at least one impairing substance. Alcohol was detected in 38% of drivers, and 69% of those drivers were above the legal blood alcohol concentration (BAC) limit of 0.08% BAC. THC was detected in 17% of drivers. Central nervous system (CNS) depressants were detected in 31%, opioids in 14%, and CNS stimulants in 13% of drivers. About a third (32%) of drivers tested positive for more than one substance. Conclusion: These findings reveal that a large proportion of injured off-road vehicle drivers test positive for substances, which indicates a serious public health issue. Targeted interventions for at-risk populations and increased law enforcement are required to reduce impaired driving among off-road vehicle drivers.
Access status: Open Access ,
UNSOURCED RANDOM ACCESS WITH USER LOCALIZATION AND SENSING
(2025-12-14) Soltani, Roshanak; Yes; Doctor of Philosophy; Department of Electrical & Computer Engineering; Not Applicable; Dr. Ekram Hossain; Yes; Dr. Jason Gu; Dr. Alex Brodsky; Dr. Dmitry Trukhachev
Future wireless networks must support massive numbers of sporadic transmissions while meeting strict spectrum and energy limits. Unsourced random access (URA) provides a grant-free model in which devices transmit without pre-allocated resources and user identifiers and the receiver jointly detects user activity and decodes messages from a large number of users. This thesis advances URA by enabling sensing and localization within the communication process to build scalable frameworks that remain reliable and energy-efficient under heavy loads. It also contributes URA designs that tolerate relaxed synchronization and support longer payloads when needed. Amulti-antenna preamble–payloadURAarchitecture is developed that fuses compressedsensing (CS)-based activity detection with iterative multiuser detection (MUD), decoding, and channel refinement on Rayleigh fading channels. The design is extended to lowscattering regimes through channel models that account for array geometry, consistent with high-frequency environments. Building on this foundation, the thesis applies integrated sensing and communication (ISAC) principles. A location-based URA framework is introduced in which the base station (BS) localizes active users while detecting and decoding their data using only uplink signals. The method ties user transmission features to angle-ofarrival (AoA) and partitions space into sectors with sub-pool allocation and reuse to reduce collisions and complexity. Localization improves communication by providing spatial information that helps separate multiple packets. The concept is further generalized in the random-access procedure. The thesis also demonstrates environment sensing where user uplink signals act as opportunistic illuminators for multi-object sensing at the BS, revealing the synergy and trade-offs between sensing and communication. Some structural challenges of URA are also addressed. Scalability under timing uncertainty is achieved through a fully asynchronous URA design that removes slot and beacon requirements and performs sliding window timing acquisition together with MUD on the Gaussian channel. For longer messages a multi-segment URA structure is proposed that stitches decoded segments across consecutive slots and employs a dual preamble pool to identify the first segment and reduce collisions. Together, these studies show how URA can jointly deliver reliable data detection, user localization, and environmental sensing with high scalability, paving the way for intelligent and spectrum efficient random-access in future networks.
Access status: Embargo ,
A TRIPLE-MODALITY STRATEGY OF NKT CELL IMMUNOTHERAPY, ONCOLYTIC VIROTHERAPY, AND CHECKPOINT BLOCKADE FOR LUNG CANCER
(2025-12-15) Lukacs, Jordan; Not Applicable; Doctor of Philosophy; Department of Microbiology & Immunology; Not Applicable; Dr. Mansour Haeryfar; Not Applicable; Dr. Jeanette Boudreau; Dr. Graham Dellaire; Dr. Roy Duncan; Dr. Brent Johnston
Lung cancer is the leading cause of cancer-related deaths in Canada. Current standard therapies, such as chemotherapy, radiation therapy, or immune checkpoint blockade (ICB) are often ineffective on their own due to severe adverse effects and acquired drug resistance. Therefore, new treatments that are safer and more effective are needed. Recent clinical trials combining ICB with other therapies have demonstrated durable outcomes in some patients with lung cancer. Here I examined the efficacy of combining PD-1 checkpoint blockade with natural killer T (NKT) cell activation therapy and recombinant oncolytic vesicular stomatitis virus (VSV-ΔM51) expressing cytokines IL-12 (VSV-IL-12) or IL-15 (VSV-IL-15), or fusion- associated small transmembrane (FAST) proteins p14 (VSV-p14), p15 (VSV-p15), and p14endp15 (VSV-p14endp15). VSV-p14, VSV-p15, and VSV-p14endp15 demonstrated enhanced immunogenic cell death and killing capacity relative to the parental virus (VSV-GFP) in lung cancer cells, in vitro. Furthermore, VSV-FAST constructs induced PANoptosis and overcame the blockade of multiple programmed cell death pathways to effectively eliminate lung cancer cells. In a genetic mouse model of lung adenocarcinoma, the combination of VSV-p14, VSV-p15, or VSV-p14endp15 with NKT immunotherapy increased overall survival relative to untreated mice. Addition of PD-1 blockade to NKT immunotherapy and VSV-FAST constructs significantly extended the survival over untreated mice. Mixed delivery of VSV-p15 and VSV- IL-12, along with NKT immunotherapy and PD-1 inhibition, had the greatest increase in overall survival of all tested combinations. Maintenance therapy with multiple rounds of VSV-p15/NKT activation/PD-1 blockade also led to significantly enhanced overall survival rates in this model, however some female mice experienced adverse reactions to the treatment. Despite increases in overall survival and increased numbers of NKT cells in the lung and spleen, there were no significant differences in total tumor area or total number of other immune cell populations within the lungs of treated mice compared to untreated mice. We also observed increased VSV neutralizing activity in the serum of treated mice, suggesting potential challenges with repeated systemic VSV treatments. Collectively, our findings demonstrate that the combination of PD-1 blockade with NKT cell immunotherapy and oncolytic VSV-p15/VSV-IL-12 presents a promising strategy for the treatment of lung cancer.
Access status: Open Access ,
Spatial and Temporal Mechanisms Controlling Convection Over The Great Plains
(2025-12-15) Verevkin, Iaroslav; Not Applicable; Doctor of Philosophy; Department of Physics & Atmospheric Science; Not Applicable; Yanping Li; Yes; Thomas Duck; Glen Lesins; Ian Folkins
The central United States exhibits an anomalous summertime diurnal cycle of precipitation, characterized by an afternoon maximum over the Rocky Mountains that transitions to a nocturnal maximum over the Great Plains. This phenomenon, which remains a challenge for many numerical models, is governed by the interaction of processes spanning multiple scales. This thesis investigates the spatial, temporal, and dynamical mechanisms controlling this diurnal cycle through a comprehensive, climatologically-grounded multi-year analysis of satellite-derived precipitation data Integrated Multi-satellitE Retrievals for GPM (IMERG) and hourly meteorological analyses Rapid Refresh and Rapid Update Cycle (RAP/RUC). The research is presented in three parts. First, the spatial variation in the synoptic structure of convective systems is examined. The analysis reveals a distinct transition in dominant forcing mechanisms with distance from the mountains: convection in the “Near Plains” (west of 100◦W) is significantly influenced by mountain-initiated solenoidal circulations, while convection in the “Far Plains” is more closely associated with the dynamics of the Great Plains Low-Level Jet (GPLLJ). Second, the thesis investigates the diurnal evolution of vertical profiles of convection. A systematic diurnal shift from surface-based to elevated convection is identified, which consistently occurs as the nocturnal boundary layer stabilizes. This shift is linked to a threshold in the low-level lapse rate of approximately -4 to -5 K/km, providing a quantifiable metric for the influence of boundary layer thermodynamics on the convective mode. Finally, the thesis examines the climatological eastward propagation of rainfall. The analysis demonstrates that the diurnal, clockwise rotation of the GPLLJ’s wind vector drives a propagating pattern of low-level mass convergence across the plains. This mechanism is modulated by topographically-induced suppression of afternoon convection via enhanced convective inhibition (CIN), enabling the nocturnal, dynamicallydriven rainfall maximum to dominate.
Access status: Open Access ,
UTILIZING MACHINE LEARNING TO DETECT TOR TRAFFIC: A REALISTIC DATASET AND A PRELIMINARY ANALYSIS
(2025-12-15) Sadik, Md Rafiqul Islam; Not Applicable; Master of Computer Science; Faculty of Computer Science; Not Applicable; Dr. Xichen Zhang; Not Applicable; Dr. Samer Lahoud; Dr. Qiang Ye
With the increasing use of anonymization technologies such as the Tor network, the ability to accurately differentiate Tor traffic from conventional Internet traffic has become an important challenge for network analysis and security monitoring.This thesis presents a fully controlled and reproducible framework for generating realistic Tor and Non-Tor traffic datasets to support the evaluation of encrypted traffic detection techniques. The framework integrates a Debian workstation, a Whonix gateway for Tor routing, and a noise-free AWS-based web server, combined with Selenium-driven automation to execute identical user activities over both Tor-based and Non-Tor-based network paths.Using this environment, a comprehensive dataset was generated across six application categories: web browsing, video streaming, file transfer, instant messaging, voice over IP, and video conferencing. A set of six machine-learning models-Decision Tree, Random Forest, XGBoost, Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) was evaluated on the generated dataset. Experimental results demonstrate that traditional treebased models, particularly Random Forest and XGBoost, consistently outperform deep-learning approaches, achieving high detection accuracy in distinguishing Tor from Non-Tor network flows across all traffic types.These findings highlight both the effectiveness of classical machinelearning approaches and the importance of realistic dataset generation in advancing encrypted traffic classification research.
Access status: Open Access ,
Continuous Ceilometer-Derived Mixed Layer Height Observations at an Eddy Covariance Flux Tower Site in the Semi-Arid U.S. Southwest
(2025-12-15) Mengering, Deklan; No; Master of Science; Department of Physics & Atmospheric Science; Not Applicable; n/a; Not Applicable; Rachel Chang; Daniel Nadeau; Manuel Helbig
The planetary boundary layer (PBL) is the atmospheric layer directly influenced by the Earth’s surface. The height of the top of the PBL (PBLH) is a key parameter that determines the extent of vertical mixing, which plays a crucial role in micrometeorology, pollution dispersion, and the exchange of energy and matter between the land and the atmosphere. In the past, PBLH estimates have relied largely on observations with low temporal resolution (radiosondes) or on modelling outputs (reanalysis). However, growing interest in the integration of PBLH measurements at surface energy flux tower sites has highlighted the temporal, spatial, and financial trade-offs that come with different PBL height measurement techniques. Of the available techniques, ceilometers offer a promising solution, as they have the potential for wide-range integration at a comparatively low cost. However, there is no publicly available, cross-platform algorithm for the detection of the PBL using ceilometer data. In this study, we developed a Python-based algorithm that can be applied to Lidar backscatter profiles from different ceilometer types. We use this algorithm to create a 5-year time series of PBLH estimates from a semi-arid high desert ecosystem in Arizona. The ceilometer PBLH estimates were compared to PBLH estimates using three independent approaches: reanalysis data, nearby National Weather Service radiosonde observations, and on-site field radiosonde campaigns. We use these estimates to validate the PBL heights derived using the newly developed algorithm. The time series of PBLH measurements is then used in combination with surface energy flux observations to investigate the relationship between PBL growth dynamics and the surface energy balance measured at a nearby flux tower. The maximum daily mean PBLH occurred in June, reaching 3100 m when the median Bowen ratio for the month was 2.6. The PBLH then decreased by 200 m in July, and another 300 m in August when the Bowen ratios dropped to 0.7 and 0.2, respectively. The results show that strong land-atmosphere coupling is characteristic for the site, with both the terrestrial (i.e. soil moisture to surface flux) and the atmospheric (i.e., surface flux to boundary layer atmosphere) legs playing an important role for the seasonal dynamics of PBLH. This study demonstrates how ceilometer-derived PBLH estimates at surface energy flux tower sites can be used to improve the current understanding of land-atmosphere coupling.
Access status: Open Access ,
UNDERSTANDING THE EXPERIENCE OF PHYSICAL DISABILITY AMONG PEOPLE LIVING WITH HIV: A QUALITATIVE STUDY
(2025-12-16) Shahzadi, Anita; No; Master of Science; School of Physiotherapy (Rehabilitation Research); Received; na; No; Caitlin McArthur; Kelly O'Brien; Lisa Barrett; Sorayya Askari; Adria Quigley
Since the advent of antiretroviral therapy, people living with HIV (PLHIV) are living longer but increasingly experience physical disability. The primary objective of this study was to understand the experience (presence and impact) of physical disability among PLHIV. The secondary objective was to understand the influence of extrinsic and intrinsic contextual factors. Interview guide developed using the Episodic Disability Framework and items from existing physical performance measures. Semi-structured interviews conducted online were audio-recorded, transcribed verbatim, and analyzed using hybrid inductive-deductive thematic analysis. Fifteen participants completed the World Health Organization Disability Assessment Schedule (WHODAS 2.0), indicating moderate disability, while interviews revealed seven themes: physical symptoms and impairments; reconstituting day-to-day activities; social disconnection and relationship changes; the episodic nature of disability and uncertainty; intrinsic strategies for resilience and adaptation; extrinsic unmet rehabilitation and service needs; and triggers. The findings from this study contribute to understanding the experience of physical disability among PLHIV.
Access status: Embargo ,
Detecting Influenza A Viruses in Surface Water: A Comparison of Passive Sampling to USEPA Method 1615
(2025-12-16) Prest, Taylor; Not Applicable; Master of Applied Science; Department of Civil and Resource Engineering; Not Applicable; n/a; Not Applicable; Amina Stoddart; Gordan Price; Graham Gagnon
Climate change, urbanization, and agricultural intensification are increasing the risk of viral contamination in surface waters, which act as reservoirs and transmission pathways for waterborne viruses. Influenza A (INFA), specifically, avian influenza strains, carried primarily by wild waterfowl, pose significant threats to animal, human, and economic health, making freshwater environments optimal for early surveillance. This study evaluated the performance of granular activated carbon (GAC)-based passive samplers paired with reverse transcription quantitative polymerase chain reaction (RT-qPCR) for detecting INFA, avian influenza H5, and pepper mild mottle virus (PMMoV) genes in lakes and rivers across Nova Scotia. Passive samplers were deployed weekly at four lakes and one river, with INFA concentrations ranging from 1.7 × 104 to 5.6 × 1012 GC sampler-1 and H5 from 6.2 × 104 to 1.8 × 1011 GC sampler-1. PMMoV was detected at two sites. Detection frequency corresponded with fall migration and waterfowl presence, while emphasizing the influence of wildlife, agriculture, and wastewater inputs. When compared to an adapted USEPA Method 1615 (electropositive filtration), GAC based passive sampling displayed comparable INFA detection, detected H5 where filtration did not, and missed PMMoV detections captured by filtration. Field and bench-scale analyses of NanoCeram filtration indicated no significant difference in detection across tested flow rates (4, 7, 10 LPM) and volumes (84.6 to 360 L). These findings highlighted the strong potential of GAC-based passive sampling as a feasible and complementary tool for avian influenza surveillance in freshwater systems.
Access status: Open Access ,
HEALTH SYSTEM READINESS TO ADDRESS THE NEEDS OF SURVIVORS OF DOMESTIC AND SEXUALIZED VIOLENCE DURING AND BEYOND PREGNANCY: A MIXED-METHODS STUDY OF SURVIVORS AND REPRODUCTIVE HEALTHCARE PROVIDERS IN NOVA SCOTIA, CANADA
(2025-12-15) Cullum, Jessie; No; Master of Science; Department of Community Health & Epidemiology; Not Applicable; NA; No; Michelle Dimitris; Jill Hayden; George Kephart; Alexa Yakubovich
Background: Intimate partner violence, the most common form of violence against women, was declared an epidemic in Nova Scotia in September 2024. The perinatal period is a time of heightened vulnerability to IPV and its health impacts, making reproductive healthcare (RHC) an opportune intervention point. However, the capacity to manage VAW across RHC in Nova Scotia is unknown. Objectives: (1) To describe the health status, experiences of violence, and characteristics of service access among survivors of domestic and sexualized violence (DV/SV) with recent pregnancies across Eastern Canada, and (2) to describe the readiness (including knowledge, opinions, and practices) of RHC professionals (RHCPs) in Nova Scotia to manage DV/SV. Methods: This convergent mixed-methods study used data from The Interprovincial Violence Against Women Project, which conducted online surveys with survivors who sought supportive services since the onset of the COVID-19 pandemic in March 2020 and health professionals working across Nova Scotia. Two study samples and subsamples were included in the present study: (1) 188 survivors, including 63 survivors with recent pregnancies; and (2) 1,649 healthcare providers, including 206 RHCPs. I descriptively analyzed quantitative data, applied reflexive thematic analysis to qualitative data, and then integrated each strand to address each of my objectives. Results: Survivors with recent pregnancies reported increased exposure to physical IPV and substantial barriers to care, including fear of police and child protective services, which were expanded upon in the first qualitative theme “control exerted by justice-related bodies.” They often reported chronic illness and heightened vulnerability to physical IPV, which the second qualitative theme, “the impact of COVID-19 restrictions on survivor-mothers” demonstrated were intensified during the COVID-19 pandemic. RHCPs reported limited preparedness to assess the safety of a survivor’s child, which contrasted the third theme, “centering the family and child impacts of DV”, where child safety was highlighted as a priority of their role. Lastly, they also reported a limited capacity to address DV/SV among patients, which the final qualitative theme “responding to perinatal DV/SV through a trauma-informed lens” contextualized as a lack of capacity to provide trauma and violence-informed care, a key strategy used by RHCPs to address DV/SV. Conclusions: This thesis is the first Canadian mixed-methods study to investigate both the needs of perinatal survivors and the capacity of RHC to attend to these needs. Findings reinforce the unique needs of survivors in and around pregnancy, including experiences of more severe intimate partner violence and considerable barriers to care, often centred around inadequate childcare support and fear of justice involvement. RHC should adapt intervention strategies for addressing DV/SV among patients that attends to these unique needs, including tailored training for providers and appropriate resourcing for safe and effective identification and response, including referral supports.
Access status: Open Access ,
EFFECT OF MICRONIZED COPPER AZOLE (MCA) ON FUNGAL AND DECAY RESISTANCE OF CANADIAN TIMBER SPECIES
(2025-12-13) Sullivan, Erin; Not Applicable; Master of Science; Faculty of Agriculture; Not Applicable; NA; No; Dr. Kimberly Pennie; Dr. Lord Abbey; Dr. Sophia He; Dr. Chijioke Emenike
The efficacy of the wood preservative micronized copper azole (MCA) was evaluated using two fungal studies: a bioassay plate study and a soil/block study. Five fungal species and six Canadian wood species were used for the experiments. The bioassay study tested all fungal species against ten different treatments. The soil/block study examined both fungal and wood species at nine different treatments. The bioassay revealed MCA concentrations of 1.5% and 2.0% were most effective at inhibiting fungi growth on petri plates. While the MicroShades natural brown (MSNB) additive dye yielded no significant effect on preservative performance. The soil/block study found that Canadian wood species treated with MCA were able to perform equally, and in some cases better, than the industry standard chromated copper arsenate (CCA). Results were concentration-specific and fungi-dependant for many wood species. Overall, MCA-treated Canadian wood species are effective in preventing mass loss against wood-rot fungi.
Access status: Open Access ,
EXPLORING LEARNING WITHIN A DECOLONIAL CONSERVATION PARTNERSHIP: IMPACTS ON EVOLVING RELATIONSHIPS AND RECONCILIATION IN A CANADIAN CONTEXT
(2025-12-15) Boardman, Sophie; Not Applicable; Master of Environmental Studies; School for Resource & Environmental Studies; Received; Alana Westwood; No; Noella Gray; Melanie Zurba
Conservation partnerships are projects with various goals around management of protected areas, species, capacity building, education, or financial and technical support that can involve Indigenous Peoples, non-government organizations, the Canadian state, the public, and researchers. Conservation partnerships are becoming increasingly interested in respecting and affirming Indigenous Peoples’ rights and activities that promote Indigenous Peoples’ reconnection and relationship to their territories. Relational aspects, such as trust, power dynamics, and how conflict is handled, are important for partnerships to be effective. As Indigenous-led conservation initiatives increase and efforts toward Reconciliation are made, the relationships of those involved continue to change, thus, learning around relationships must be examined. Social learning theory is a useful tool to examine learning since it focuses on what was learned and how learning happened. This study focused on the Conservation through Reconciliation Partnership (CRP) as a case study to examine what kinds of shifts in relationships are taking place as this partnership embodies learning and generating knowledge to support Indigenous-led conservation. This study used social learning to explore the learning outcomes and processes of relationships and relationship-building of those involved with the CRP. The following objectives guided this research: (1) identify the SL processes and outcomes that emerged from relationships developed among those individuals involved with the CRP, and (2) explore if and how individuals’ learning outcomes from engagement with the CRP inform relationship-building practices outside the CRP, and (3) explore how evolving relationships between CRP participants are impacting personal and organizational understandings and actions of reconciliation. Fourteen people engaged with the CRP were interviewed following a semi-structured format. The results found many relationship outcomes and processes that allowed the CRP to be a positive and collaborative space and combined with the relational aspects of strong leadership and facilitation, which allowed trust to be fostered throughout engagements. Only a few participants’ learning outcomes were brought back to their organizations due to power, such as job role and who is perceived as being able to change social norms. Participants also shared diverse perspectives about reconciliation and what changes are needed on a personal, organizational, and societal level. Many shared the importance of relationship building within the CRP as key to helping them learn or recognize spaces where reconciliation may not be as accepted. A few participants shared criticisms about reconciliation and the colonial undertones of the term and conservation might undermine relationship efforts. These findings demonstrate the importance of examining the relational aspects of partnerships so they can be equitable and as reconciliation continues to be advanced within Canada.
Access status: Open Access ,
Generalizing the Linear Step-up Procedure for False Discovery Rate Control with Applications to Setwise and High Dimensional Variable Selection
(2025-12-15) Organ, Sarah; Not Applicable; Doctor of Philosophy; Department of Mathematics & Statistics - Statistics Division; Not Applicable; Jiahua Chen; Not Applicable; Edward Susko; Lam Ho; Hong Gu; Toby Kenney
This thesis presents a unified framework for false discovery rate (FDR) controlled variable selection that addresses three major challenges: (1) the inability of traditional FDR procedures to accommodate structured, non-independent hypothesis testing, (2) the failure of standard FDR control procedures for variable selection in the presence of strong dependence among predictor variables, and (3) the lack of methodology for obtaining valid p-values for applying FDR control in the high-dimensional data where the number of variables exceeds the number of samples (m > n). To address the first challenge, we generalize the linear step-up procedure by applying a sizing function to account for dependence structures among hypotheses. This involves a theoretical extension of the linear step-up framework to structured collections of dependent hypotheses, ensuring that FDR control is preserved under dependence when applied in structured hypothesis settings. We provide theoretical guarantees of this generalization for FDR control. These generalized procedures form the foundation for both SHRED and HVS methods we later describe. To address the second challenge, we propose a family of methods for setwise variable selection that extend the classical FDR framework to non-independent hypothesis spaces. We first introduce the Setwise Hierarchical Rate of Erroneous Discovery (SHRED) methods, which performs FDR-controlled variable selection over hierarchical trees of hypotheses. For these methods, we expand on the notion of variable selection such that we allow for selecting sets of highly correlated surrogate variables, under the assumption that at least one variable in the set is a true variable. We then introduce the Hypergraph Variable Selection (HVS) method, which enables setwise variable selection over a hypergraph of hypotheses representing complex dependencies among variables. In simulation studies we show that both the SHRED and HVS methods have significant advantages over current FDR control methods for variable selection when there is correlation present among predictor variables. Finally, to overcome the third challenge, the lack of methodology for obtaining valid p-values in high-dimensional settings, we introduce a multi-step p-value estimation procedure that enables the application of linear step-up methods for FDR control when m > n. This procedure begins with a data reduction step to identify a subset of variables believed to contain all true signals, followed by a conservative re-fitting step that yields valid p-values for inference. We provide theoretical guarantees showing that this approach preserves FDR control, as well as simulation studies highlighting the advantages of this method over classic high dimensional variable selection methods for FDR control. Collectively, these contributions advance the theory and practice of FDR-controlled variable selection, offering a flexible and principled framework that accommodates structure, dependence, and high dimensionality, three fundamental challenges of variable selection with FDR control.
Access status: Open Access ,
Representing Gait Outcomes For Advanced Knee Osteoarthritis Using a Wearable Inertial Sensor System
(2025-12-15) MacDonald, Benjamin; No; Master of Applied Science; School of Biomedical Engineering; Received; Dr. Derek Rutherford; No; Dr. Scott Landry; Dr. Cheryl Kozey; Dr. Janie Astephen Wilson; Dr. Michael Dunbar
This thesis aimed to develop an efficient Inertial Measurement Unit (IMU) protocol for capturing kinematic and kinetic gait patterns in knee Osteoarthritis (OA) patients in a free-living environment, moving beyond costly lab-based motion capture. The study involved end-stage knee OA patients before and after knee arthroplasty (KA). The first objective showed good statistical agreement between IMU-derived and motion capture discrete metrics for knee adduction and flexion angles, with differences within clinically acceptable limits (minimum clinically important difference or typical inter-session variability). The second objective explored using IMU-derived features (accelerations/angular velocities) to estimate the temporal patterns of kinetic waveforms (flexion/adduction moments). Shank-only models explained over half the variance, improving with foot sensor data (R2=0.55-0.69). These results validate the ability to capture clinically relevant gait outcomes with IMU's, enabling the transition to remote, continuous monitoring of OA progression and recovery outside of the clinic.
Access status: Open Access ,
Exploring Patterns of Emergency Department Use for People Experiencing Homelessness in New Brunswick
(2025-12-15) Jenna, Hepburn; No; Master of Science; Department of Community Health & Epidemiology; Received; n/a; Not Applicable; Leslie Anne Campbell; Kavish Chandra; Daniel Dutton
Background: Homelessness is a complex and growing issue that negatively affects both individuals and communities. People experiencing homelessness (PEH) often have high levels of unmet health care needs resulting from higher risk of poor health and greater challenges in accessing appropriate health care than the general population. In Atlantic Canada, housing prices are rapidly increasing, reducing housing affordability and accessibility. These population-level factors may play a role in the health of PEH. Objectives: 1) Among individuals with visits to the Emergency Department at Horizon Health, identify PEH and people not experiencing homelessness (“controls”) and describe their demographics, reasons for using the ED, triage level, number of ED visits, and whether they have a primary care provider; 2) To compare ED use among PEH and controls to determine whether there is a difference in the total number of visits, triage level, and whether they have a primary care provider; 3) To determine whether changing housing prices are associated with differences in ED presentation (number of visits and/or reasons) for PEH and controls between 2020 and 2024 across three cities (Fredericton, Moncton, and Saint John). Methods: A repeated cross-sectional study was conducted using administrative health data from four hospitals across three NB cities. Patients were defined as homeless if they were identified as experiencing homelessness in their record or give the address of a shelter or hotel. Nearest neighbour propensity score matching was used to establish a comparable control group, matching patients on facility, sex, and age. Generalized linear models (GLM) were used to assess the effect of experiencing homelessness on the total number of visits (Poisson), average triage level (linear), and primary care provider status (logistic) of PEH and matched controls. Rental price and vacancy rates were graphed by city and year, and correlation coefficients are estimated against ED visits. Results: Between 2020 and 2024, there were 810 PEH and 226,729 housed individuals who presented to the ED. Compared to matched controls, the rate of ED use for PEH was 1.55 times greater for PEH than matched controls (IRR = 1.55, CI: [1.45, 1.66], p < 0.001). Mental health presentations were four times more common among PEH (18.5%) than among controls (4.7%). PEH have 57.7% lower odds of having a primary care provider compared to controls (OR = 0.42, CI: [0.34, 0.52], p < 0.001). PEH had an average 0.11 point lower average CTAS score (coef. = -0.11, CI: [-0.18, -0.03], p = 0.004). ED visits by PEH increased in all cities over the study period, with a particularly sharp rise in Saint John. Rental prices also increased and correlated positively between ED visits by PEH and lowest quartile rental prices (r = 0.36 for one-bedroom units), and negatively with vacancy rate (r = -0.27). Conclusion: This study provides one of the first population-level analyses of ED use among PEH in NB. PEH have higher ED utilization, more acute presentations, and lower access to primary care compared to housed individuals. Patterns of ED use appeared to reflect broader trends in housing affordability, highlighting the need for coordinated housing and healthcare policy interventions.
Access status: Open Access ,
UNCERTAINTY AWARE TEMPORAL RELATIONSHIP MODELLING IN KNOWLEDGE GRAPHS
(2025-12-15) Ramachandran, Deepa; Not Applicable; Master of Computer Science; Faculty of Computer Science; Not Applicable; na; Not Applicable; Dr. Evangelos Milios; Dr. Gabriel Spadon De Souza; Dr. Ga Wu; Dr. Frank Rudzicz
Existing approaches for constructing knowledge graphs (KGs) from text sources effectively capture structural relationships but often struggle to represent the temporal uncertainty inherent in event-based information. Their temporal encoding typically relies on exact timestamp matching, which limits the ability to handle natural temporal variation among events. This thesis introduces Probabilistic Temporal Relationship Modeling (PTRM), a framework that represents events as temporal distributions to capture uncertainty in timing and relationship propagation. PTRM employs distributional embeddings that jointly encode semantic and temporal uncertainty, enabling soft temporal matching through probabilistic overlap computation. Experiments involving 200,000 events from the ICEWS and GDELT datasets demonstrate that PTRM substantially outperforms existing baselines, including temporal knowledge graph embedding methods and classical approaches, while maintaining strong calibration properties essential for decision-critical applications. The results demonstrate that PTRM bridges the gap between rigid timestamp-based methods and the temporal flexibility required for real-world reasoning, achieving state-of-the-art performance in temporal relationship inference under uncertainty.
Access status: Open Access ,
PUBLICLY FUNDED GUIDELINE-RECOMMENDED MEDICATIONS DISPENSED TO COMMUNITY-DWELLING OLDER ADULTS IN NEW BRUNSWICK FOLLOWING HOSPITAL DISCHARGE FOR MYOCARDIAL INFARCTION BETWEEN 2009 AND 2017
(2025-12-11) Manning, Devin; No; Master of Science; Department of Community Health & Epidemiology; Received; na; No; Dr. Sanja Stanojevic; Dr. Rob Stevenson; Dr. Daniel Dutton; Dr. Adrian Levy
BACKGROUND: Pharmacologic therapy plays a central role in secondary prevention following myocardial infarction (MI). Evidence from randomized controlled trials has demonstrated that β-blockers, statins, renin–angiotensin–aldosterone system (RAAS) inhibitors, and antiplatelet agents, particularly P2Y12 inhibitors, reduce mortality, recurrent infarction, and stroke post-MI. Studies in Canada have shown that their use in routine practice remains variable and have not examined the determinants of medication use since P2Y12 inhibitors became standard of care. In New Brunswick (NB), population-based evidence is limited. Understanding these variations is critical for assessing the quality of post-MI care and informing system-level improvements. OBJECTIVES: This study aimed (1) to calculate the annual proportion of community-dwelling older adults (66 years or older) in NB with publicly funded drug insurance and received a dispensation for each guideline-recommended medication class: β-blockers, statins, RAAS agents, and P2Y12 inhibitors, within 90 days of discharge for MI between 2009 and 2017, and determine whether this changed over time; and (2) to determine whether characteristics of the discharging physician were associated with the likelihood of an individual receiving a P2Y12 inhibitor within 90 days of discharge. METHODS: A retrospective, population-based cohort study was conducted using linked administrative health data. The cohort included community-dwelling older adults who were discharged alive from hospital with a primary diagnosis of MI (ICD-10 I21) and had public drug coverage during the study period. Dispensation of each recommended medication within 90 days of discharge was identified from outpatient pharmacy claims. Logistic regression models with generalized estimating equations were used to estimate the associations between physician characteristics and P2Y12 inhibitor use, adjusting for patient age, sex, and length of hospital stay. RESULTS: Among the 2,674 eligible patients, the median age was 78 years (IQR: 71 to 84), and 47% were female. Overall, 79% of patients received a β-blocker, 83% a statin, 72% a RAAS inhibitor, and 70% a P2Y12 inhibitor within 90 days of discharge. Dispensation rate for β-blockers, statins, and P2Y12 inhibitors remained stable across the study period, while RAAS inhibitor use declined. Physician-level variation was evident, compared to cardiologists, patients discharged by general practitioners (OR 0.38; 95% CI 0.28, 0.51) or other specialists (OR 0.47; 95% CI 0.32, 0.68) were less likely to receive P2Y12 inhibitors post-discharge. Patients discharged by female physicians were less likely to receive P2Y12 inhibitors (OR 0.69; 95% CI 0.51, 0.92). No statistically significant associations were found with physician graduation decade or training location. DISCUSSION: Post-MI use of evidence-based pharmacotherapies among publicly insured older adults in NB was comparable to rates reported elsewhere in Canada and remained below recommended targets. The observed variation in P2Y12 inhibitor use across physician groups suggest that structural factors, such as clinical specialization, and drug reimbursement policies, may influence access to guideline-recommended therapies.
Access status: Open Access ,
Integrating DVFS and Task Scheduling to Improve Energy Efficiency for Heterogeneous Edge Devices: A Reinforcement Learning Approach
(2025-12-13) Wang, Haoyu; Not Applicable; Master of Computer Science; Faculty of Computer Science; Not Applicable; n/a; Not Applicable; Saurabh Dey; Jie Gao; Qiang Ye; Man Lin
Energy efficiency is a primary design objective for embedded and edge computing platforms, which operate under tight power and thermal constraints while serving latency-sensitive workloads. In this thesis, we focus on CPU power management on heterogeneous big.LITTLE systems for single-threaded, periodic tasks that operate under a soft Target Execution Time (TET) constraint. Specifically, we design and implement a user-space, reinforcement-learning-based controller that jointly performs dynamic voltage and frequency scaling (DVFS) and task scheduling on heterogeneous edge devices. The controller uses a learned policy to select both the CPU cluster and operating performance point in each execution window so as to minimize per-window energy consumption while satisfying TET constraints. A compact, time-aware state representation makes the policy explicitly TET-conditioned, enabling it to adapt to different TET values at runtime without retraining. Using a hardware-in-the-loop evaluation on an ODROID-N2+ platform, we compare the learned policy against the standard Linux \texttt{ondemand} governor on keyword spotting (KWS) and YOLO-lite object detection workloads. With TETs randomly drawn from the 3.5--4.5\,s range, the proposed controller reduces per-cycle energy by up to 10.6\% for KWS and 7.3\% for YOLO-lite while maintaining high TET satisfaction rates.
Access status: Open Access ,
'And that one takes a big bite of one of those nice red apples': Portraits of Native Women in Thomas King's Green Grass. Running Water and Medicine River
(1998-04) McKay, Christina A.S.; Not Applicable; Master of Arts; Department of English; Not Applicable; unknown; Not Applicable; Patricia Monk; R. Tetreault; J.A. Wainwright
Given the multiplicity of voices and perspectives (female and male, Native and non-Native) existing in Canada, the question now is how we might collectively learn to reconcile our differences. In western culture, both gender and culture are thought to be fixed, predetermined and separate. But when viewed from a Native perspective, gender and culture are actually aspectival, rather than essential. The women in Thomas King's novels Green Grass, Running Water and Medicine River shed light on a Native world view which resists white western patriarchal assumptions about culture and gender and suggests entirely different roles for women and a new paradigm for human relationships. In Chapter One, I explore how and why the essentialized image of the "Imaginary Indian" arose and persists in white western culture and the consequences this stereotype has for Native peoples, and Native women in particular. In Chapter Two, I examine the ways both the Native and non-Native women in GGRW and MR are marginalized by the white western culture's essential views about race and gender. The women in King's novels prove their capacity to resist racism and sexism and survive by showing their communities how to look at the world from multiple perspectives. Chapter Three looks at the strategies Native women use not only to resist maginalization, but to re-imagine and transform the culture(s) they live in. These strategies are based on the Native practice of braiding. In the same way that different strands weave together to form a braid, culture and gender are also not individualistic (essential), but individualized (aspectival). My conclusion will extend my discussion of Thomas King's writing to address the question of cultural difference and Native values in Canada. I will make particular reference to the current debate about Canadian immigration policy, which, rather than embracing difference, suggests that immigrants must conform to the cultural norms of the dominant society in Canada.
Access status: Open Access ,
Neural Compression for Scalable Question-Answer Retrieval
(2025-12-11) Khiet, Moamen; Yes; Master of Computer Science; Faculty of Computer Science; Not Applicable; n/a; Not Applicable; Dr. Hassan Sajjad; Dr. Frank Rudzicz; Dr. Ga Wu
Question-answering systems at scale face fundamental performance barriers when traditional vector databases transition from exact to approximate search, causing substantial degradation in both query throughput and retrieval quality. While compression can address these challenges, existing compression approaches either apply generic transformations ignoring retrieval task structure (PCA) or require retraining entire embedding models (Matryoshka), limiting practical applicability. This thesis introduces neural compression for question-answer retrieval through two-stage learning that compresses 384-dimensional context embeddings to 32 or 64 dimensions while preserving semantic information. The approach trains an autoencoder to compress context embeddings using cosine similarity loss, then trains a mapper network to predict compressed codes directly from question embeddings using mean squared error loss in compressed space. Both networks are trained on the training split (72%, 86,400 pairs). We evaluate this approach on 120,000 question-answer pairs spanning six knowledge domains, comparing against six baseline methods (FAISS, HNSW, ScaNN, PCA, Matryoshka, zero-shot) across six dataset scales (20K to 120K) with three iterations per configuration, totaling in 198 experimental runs. Results demonstrate that neural compression achieves 0.1725 ROUGE-1 score compared to FAISS’s 0.1624 (+6.2% improvement) while reducing storage from 184 MB to 7.7 MB (96% reduction) and increasing throughput from 151 to 7,861 queries per second (52× speedup). Neural compression is the only method whose quality improves with scale (+2.5% from 20K to 120K samples) while all baselines degrade (-6% to -13%). The performance crossover occurs at approximately 40K samples, earlier than hypothesized, as FAISS quality degrades from curse of dimensionality effects before its algorithmic transition to approximation. These results show that task-specific learned compression through asymmetric architecture compressing only contexts while keeping questions full dimensional enables exact search at scales where high dimensional methods must approximate, fundamentally changing scalability characteristics of retrieval systems.
Access status: Open Access ,
Assessing the effects of sampling design and data integration in spatio-temporal fisheries models
(2025-12-11) McDonald, Raphaël; No; Doctor of Philosophy; Department of Mathematics & Statistics - Statistics Division; Not Applicable; Dvora Hart; Yes; Orla Murphy; Craig Brown; Joanna Mills Flemming; David Keith
Many communities around the world depend on the socio-cultural and economic benefits derived from fisheries. To ensure these continued benefits, fisheries management relies on accurate estimates of fish population size and health obtained through the stock assessment process. This thesis develops and improves upon various statistical methods used within the stock assessment process. We first show how common sampling designs do not strongly impact model results, but also how sampling effort allocation can introduce time-varying biases if the strata-specific sampling effort is not proportional to strata size. We next turn our attention to analyses that occur early in the stock assessment process, such as length-weight relationships. While complex statistical models are often necessary to estimate these relationships, their outputs are routinely used without accounting for their associated uncertainties. We present simple methods to highlight the increase in precision achieved by error propagation. The next two chapters focus on developing methods for the inclusion of novel types of data within stock assessment models. The first is a benthoscape map. Focusing on sea scallops in the Bay of Fundy, we demonstrate how benthoscape maps can be interpreted as habitat features and help inform the estimation of catchability and probabilities of encounter to improve parameter estimation and identify biologically meaningful area boundaries. The second is drop camera data. We propose various model modifications to account for the resolution difference between datasets. This work identifies a previously unknown relationship between the size of survey tows, the underlying spatial distribution of the population, and the accuracy of the survey. High aggregation or high noise in the spatial distribution results in the survey indices being more precise than the drop camera indices even with substantially smaller sample sizes. This thesis develops novel statistical methods useful to the broader field of fisheries science, with contributions for improved data integration, error propagation, and sampling design, while opening various new avenues of future work in all these same fields.

Browse

Recent Submissions