About MacSphere

MacSphere is McMaster University's Institutional Repository (IR). The purpose of an IR is to bring together all of a University's research under one umbrella, with an aim to preserve and provide access to that research. The research and scholarly output included in MacSphere has been selected and deposited by the individual university departments and centres on campus.

To contribute to McMaster's Institutional Repository, please sign on to MacSphere with your MAC ID.

If you have any questions, please contact the MacSphere Support Team.

Students wishing to deposit their PhD or Masters thesis, please follow the instructions outlined by the School of Graduate Studies.

Communities in MacSphere

Select a community to browse its collections.

Now showing 1 - 5 of 5

Recent Submissions

Using Machine Learning to Predict Hip Fracture Risk from Dual-Energy X-Ray Absorptiometry Images and Health Factors
(2026) Kramer, Taylor
Osteoporosis is a highly prevalent skeletal disease that greatly increases the risk of fragility fractures and affects millions of older adults worldwide. Hip fractures are especially dangerous, often resulting in long-term disability, increased need for institutional care, and increased healthcare costs. Current methods for diagnosing osteoporosis and subsequent fracture risk, such as bone mineral density (BMD) T-score from dual-energy x-ray absorptiometry (DXA) imaging and the Fracture Risk Assessment Tool (FRAX) have limitations in their ability to accurately predict fracture risk. Developments in machine learning and image processing tools have shown promise for improved fracture risk prediction. This work builds on these advances by integrating DXA imaging data with longitudinal clinical information to develop and evaluate predictive models aimed at improving fracture risk prediction. To evaluate the independent predictive ability of DXA images, a simple feed-forward neural network trained on DXA images reduced using principal component analysis (PCA) was developed. This model demonstrated that images alone contain predictive information for fracture risk; however, performance outcomes improved when clinical risk factors (CRFs) were incorporated in a multimodal feed-forward neural network approach. SHapley Additive exPlanations analysis revealed that imaging features contributed most strongly to the model’s predictions while age and body mass index (BMI) had comparatively smaller contributions. Both models achieved performance metrics within the range of values typically reported for FRAX-based assessments. Next, deep learning and transfer learning models were developed to assess how model performance was impacted by the addition of spatial pattern recognition within the images. Deep learning achieved moderate predictive ability, with performance similar to FRAX, while transfer learning models struggled to learn meaningful patterns from the available data. The purpose of this work was to improve fracture risk prediction by integrating DXA imaging data with clinical health information. The results of this work can inform the development of clinically relevant fracture risk prediction tools that support early intervention for at-risk individuals.
Adjoint-Based Inverse Design of Nanophotonic Structures for Imaging and Sensing Applications
(2026) Arfin, Rishad; Bakr, Mohamed; Electrical and Computer Engineering
This thesis proposes a systematic and efficient approach to design and optimize different classes of nanophotonic devices for emerging imaging and sensing applications. A computational inverse design approach is used to explore and discover efficient nanophotonic designs in the vast design space, achieving optimal performance. Adjoint sensitivity is highlighted and utilized in the design strategy to accelerate the development and optimization of these devices. The design methodology is demonstrated across several target applications, including complementary metal-oxide-semiconductor (CMOS) microlenses, multispectral metasurface routers for imaging, and metasurface optical sensors for gas and biosensing. The overall results presented in this thesis suggest that inverse design approaches by leveraging adjoint sensitivity provide an efficient way to develop and optimize compact nanostructures, achieving target functionalities for next-generation imaging and sensing applications.
Midwifery Services during the COVID-19 pandemic
(2026) Ku Carbonell, Susana Elsa
The COVID-19 pandemic placed unprecedented pressure on health systems worldwide, disrupting essential care services. Sexual and reproductive health services were among the most affected, yet the specific challenges faced by midwifery services remain underexplored. This dissertation examines how the pandemic affected midwifery care in Lima Metropolitana, Peru, and the Greater Toronto Area, Canada, situating these experiences within broader social, institutional, and historical contexts. Using a qualitative multiple case study design, the study employed trauma-informed, intersectional, and postcolonial feminist perspectives to interpret the relational, ethical, and structural dimensions of midwifery practice during the crisis. Findings indicate that midwives were vulnerable to fear as a manifestation of trauma, that resilience often took the form of resistance, and that the capacity to sustain care depended on access to structural supports, professional autonomy, and organizational flexibility. Core values and the philosophy of midwifery, such as relational continuity, autonomy, and respect, were both tested and upheld, while the systemic positioning of midwifery within each health system shaped these responses. The study highlights that midwifery is not only a service, but a relational and ethical practice embedded in historical and structural realities. This research contributes to knowledge by explaining how midwifery’s adaptive capacity during crises is shaped by broader structural, institutional, and social factors rather than by individual effort alone. It underscores the importance of understanding and addressing the systemic conditions that enable midwives to provide equitable, compassionate, and resilient care. By situating midwifery within the interplay of trauma, ethics, and power, this dissertation advances conceptual understanding of health service provision in crises and informs strategies to strengthen midwifery services in diverse sociopolitical contexts.
The Effect of Agrochemicals and E-waste on Adverse Pregnancy Outcomes: A Methodological Review of Systematic Reviews
(2026) Ahmed, Shakil
Background: Exposure to environmental toxicants remains a critical global public health challenge, particularly during pregnancy, when maternal and fetal systems are uniquely vulnerable. Two major and increasingly prevalent sources of environmental contamination are agrochemicals, including pesticides and chemical fertilizers and electronic waste (e-waste), which commonly includes heavy metals, brominated flame retardants, and persistent organic pollutants. A growing body of research suggests that exposure to these substances may increase the risk of adverse pregnancy outcomes (APOs). However, the methodological quality and reporting standards of systematic reviews in this area are insufficient, which detracts from the reliability of existing conclusions. This uncertainty limits the reviews' effectiveness in informing policy decisions, clinical guidance, and risk communication. Objective: This methodological review aimed to systematically identify and evaluate the methodological and reporting quality of systematic reviews and meta-analyses examining the association between exposure to agrochemicals or e-waste and adverse pregnancy outcomes. The review further sought to compare statistical synthesis approaches used across meta-analytic reviews and identify methodological gaps to inform future research practice. Methods: This review followed the PRIOR (Preferred Reporting Items for Overviews of Reviews) reporting guideline. Two PROSPERO-registered protocols were developed a priori: one focused on agrochemical exposures (CRD42024533969) and one on e-waste exposures (CRD420250627366). Eight international bibliographic databases were searched without language or date restriction. Systematic reviews and meta-analyses reporting pregnancy-related clinical outcomes following exposure to agrochemicals or e-waste were eligible. Data extraction was performed independently by two reviewers. Methodological quality was assessed using AMSTAR-2. The completeness and transparency of search strategies were evaluated using the PRESS guideline and an operationalized reporting checklist developed by Norling et al. For reviews conducting meta-analysis, statistical methodology was assessed using a structured framework based on contemporary meta-analytic best practice. Results: A total of 39 systematic reviews were included: 27 on agrochemical exposures and 12 on e-waste-related toxicants. Across both exposure domains, the majority of reviews were rated as “critically low” quality according to AMSTAR-2, largely due to recurring critical weaknesses, including lack of prospective protocol registration, incomplete justification for excluded studies, and limited incorporation of risk-of-bias assessments into interpretation. Only a small proportion of reviews met high methodological standards. Evaluation of search strategy rigour showed moderate adherence to PRESS recommendations, although use of proximity operators, controlled vocabulary expansion, and search peer review were inconsistently applied. Reporting quality of search strategies also varied substantially, with incomplete documentation of databases, search dates, citation searching, and deduplication procedures in many reviews. Meta-analytic practices were heterogeneous, with frequent use of random-effects models but limited justification for model choice and variable assessment of heterogeneity and publication bias. Conclusions: This methodological review shows that most systematic reviews examining the impacts of agrochemical and e-waste exposure on adverse pregnancy outcomes have significant methodological and reporting limitations. These weaknesses reduce confidence in existing pooled estimates and limit the interpretability of findings for clinical, research, and policy decision-making. Strengthening future evidence synthesis in this field will require routine protocol registration, comprehensive and peer-reviewed search strategies, and standardized adverse pregnancy outcome definitions.
From Exposure to Understanding: How Cognitive and Linguistic Factors Shape L2 Vocabulary Learning in Digital Environments
(2026) Gallant, Jordan; Kuperman, Victor; Cognitive Science of Language
This thesis investigates the cognitive and linguistic factors that influence second-language (L2) vocabulary and morphological learning in app-based digital environments. Across three studies, I examine how exposure type, morphological structure, semantic transparency, and learners’ L1 backgrounds shape English vocabulary acquisition and morphological abstraction. The first study focused on incidental L2 vocabulary learning within a cloze-translation task, distinguishing between target encounters (when a word is explicitly practiced) and context encounters (when it appears indirectly). Results indicated that context encounters facilitated learning even during initial target exposures, though gains from incidental learning were smaller than those from explicit practice, highlighting the role of cognitive load and task difficulty in vocabulary acquisition. The second study investigated how L2 learners acquire morphological knowledge through app-based vocabulary training, examining the roles of repetition (token frequency), variability (type frequency), and semantic transparency. Successful morphological learning depended on repeated exposure to suffixes, particularly when they appeared in diverse lexical contexts, while non-morphemic sequences conferred no advantage. Type frequency emerged as a strong predictor of learning, and L1 background modulated performance: German-speaking learners, whose L1 shares morphological and typological similarities with English, showed substantial gains, whereas Japanese-speaking learners showed minimal improvement. The third study examined compound word learning, evaluating transparency at the whole-word and constituent (modifier and head) levels. Whole-word and head transparency consistently facilitated learning across language groups, while cross-linguistic differences emerged in sensitivity to modifier transparency, suggesting an interaction between universal semantic constraints and L1-specific processing strategies. Together, these studies demonstrate that L2 vocabulary and morphological learning are experience-driven, meaning-sensitive, and shaped by linguistic background. The findings advance theoretical models of lexical processing and provide evidence-based guidance for the design of adaptive, effective digital language-learning tools.

About MacSphere

Discover

Communities in MacSphere

Recent Submissions