19614 research outputs found
Sort by
Audio analysis of customer calls for predicting purchase intentions: A novel approach to e-commerce insights
Get PDFClient audio recordings represent a valuable resource for many types of businesses. Utilizing these recordings to identify potential customers can help enhance purchase rates and reduce marketing costs, particularly with different kinds of machine learning methods that automatically label different groups, including positive, neutral, and negative buyers, instead of manual analysis. Though previous research has predominantly focused on text content analysis for this purpose, audio features, which effectively capture voice nuances such as tone, pitch, rhythm, and interaction patterns between interviewers and interviewees, may impact the model performance.
This project explored an innovative method. It firstly investigates the effectiveness of emotion detection through audio features, leveraging two datasets: the Toronto Emotional Speech Set (TESS) and the Surrey Audio-Visual Expressed Emotion Dataset (SAVEE). Furthermore, hierarchical clustering techniques are applied to explore the relationship between emotion-related audio features and customer categories using audio data provided by VINN Auto, an e-commerce firm. Next, Exploratory Data Analysis (EDA) is conducted to find the correlation between interaction-related audio features and customer categories, including positive, neutral, and negative buyers within the same dataset after labeling it. Using supervised learning, the results indicate that integrating audio features, including emotion-related and interaction pattern features, can affect the performance of models like Support Vector Machines (SVM), Decision Tree, and Extreme Gradient Boosting (XGBoosts), particularly when combined with traditional audio content-related features such as Term Frequency-Inverse Document Frequency (TF-IDF) scores while applying adjusted weight configuration for positive class. After these exploration, an ensemble method using a soft voting mechanism across these three models is developed to assess whether it can enhance the identification of potential purchasers.
The approach of combining emotion-related audio features, interaction pattern features, and content-based features like TF-IDF scores with tailored weight configurations highlights the value of collaborating audio features in customer identification tasks compared with only using content-based features like TF-IDF scores. It could be a robust strategy for improving classification outcomes for the relevant analysis in the future.Graduat
The Development of Chemical Analytical Tools for Community Drug Checking
Get PDFDrugs have many uses from pleasure to pain relief, ceremony, and medicine.
Drugs also carry risks, from uncomfortable side effects, to dependence, and even death. In the case of pharmaceuticals, most people are familiar with receiving instructional notes and warnings such as "take with food" or "let your pharmacist know about other medications that might have undesirable interactions." Strict quality control means that prescribed and regulated drug mixtures are known to be as safe as possible. In the case of the illicit drug market, such assurances of quality and support are not afforded.
In response, this thesis focuses on advancing the technology required for drug checking, a grassroots harm reduction initiative that aims to provide a level of quality control to the illicit drug market using various analytical approaches.
Due to the unprecedented and increasing number of overdose deaths, these services have been expanding throughout North America. Drug checking empowers people who use drugs with the knowledge of what they are consuming and further provides an avenue for education and support to communities about the local drug supply. However, implementation of drug checking faces many barriers not only systemically but analytically as well, in part due to the dynamic and unpredictable drug supply and demand for simple, cost-effective, and point-of-care techniques.
This thesis explores several point-of-care analytical methods in their application to drug checking. These analytical methods include immunoassay test strips, infrared, Raman, and surface enhanced Raman spectroscopy, and gas chromatography–mass spectrometry. Notably, this research and development takes place while concurrently providing drug checking as a community harm reduction service. Most of the datasets used throughout this work are acquired at the service and reflect the local drug supply. A major focus of this research is on the detection of opioids and benzodiazepines in drug mixtures. Chemometric approaches are used to evaluate, compare, and improve the capability of multiple instruments in providing useful drug information for the local supply. This includes classification and quantification schemes using a wide range of methods such as partial least squares regression, local outlier factor, principal component analysis, random forest classifier, least angle squares regression, correlation analysis, k-nearest neighbours, multivariate curve resolution, and density-based spatial clustering. Performance metrics, such as true positive rates, false positive rates, F1 scores, and receiver operating curves for qualitative detection and root mean square error and accuracy profiles for quantification, are used for evaluation. Additionally, a custom analysis platform is developed and implemented using Python and Jupyter notebooks to allow for such developments to be actualized within the service.
Beyond technical evaluation, discussion of the results of this research largely considers the practical requirements of point-of-care service delivery.
Within this work, technical information regarding drug checking technologies and data analysis is contextualized within harm reduction and contributes to strengthening the body of drug checking literature and resources.Graduate2024-12-1
A Novel Methodology to Predict the Long-Term Performance of Vacuum Insulation Panels (VIPs) Using Climate Data
Get PDFVacuum insulation panels (VIPs) have been a common insulating technology used in refrigeration and can help limit energy use in buildings by providing up to 10 times more insulation than typical insulation materials, all while using less wall space. This is specifically useful in places like Canada, where climates are cooler. Knowledge gaps around aging have currently prevented VIPs from being used in building envelope constructions. One of the remaining gaps of knowledge is that there is no methodology that has been created and linked to climate data to predict the actual performance of VIPs.
This paper starts with discussions on various factors which influence the thermal conductivity of a VIP, relates it to the climate data of Victoria, British Columbia, Canada from 1997-2021, and proposes a methodology that can predict the long-term performance of VIPs in different climates. The proposed methodology was created in a piecewise approach, starting from constant conditions of 23 ֯C and 75%RH, moving to dynamic conditions based on climate data, and then adding the presence of a getter and desiccant. The resultant methodology produced a simplistic approach that has the potential to predict the performance of VIPs in various climate conditions.
The proposed methodology shows that the thermal conductivity of VIPs remained relatively constant until either the getter or desiccant reached capacity. From there, the thermal conductivity began to increase over time. This methodology was then applied across four other (total of five) Canadian cities (Victoria, BC; Edmonton, AB; Yellowknife, NT; Ottawa, ON; Quebec City, QC), which all showed similar aging trends except for Victoria, British Columbia when reviewing ageing due to moisture content and Yellowknife, NT due to air pressure. The outputs from this methodology were also compared to the results obtained from accelerated ageing tests conducted in the laboratory, to estimate VIP parameters such as air and water vapour transmission rates, desiccant quantity, and sorption characteristics of the core material. The refined methodology can be converted into a standard method that has the potential to accurately predict VIP ageing in different climatic conditions.Graduat
Enhancing field multiplication in IoT nodes with limited resources: A low-complexity systolic array solution
Get PDFSecurity and privacy concerns pose significant obstacles to the widespread adoption of IoT technology. One potential solution to address these concerns is the implementation of cryptographic protocols on resource-constrained IoT edge nodes. However, the limited resources available on these nodes make it challenging to effectively deploy such protocols. In cryptographic systems, finite-field multiplication plays a pivotal role, with its efficiency directly impacting overall performance. To tackle these challenges, we propose an innovative and compact bit-serial systolic layout specifically designed for Montgomery multiplication in the binary-extended field. This novel multiplier structure boasts regular cell architectures and localized communication connections, making it particularly well suited for VLSI implementation. Through a comprehensive complexity analysis, our suggested design demonstrates significant improvements in both area and area–time complexities when compared to existing competitive bit-serial multiplier structures. This makes it an ideal choice for cryptographic systems operating under strict area utilization constraints, such as resource-constrained IoT nodes and tiny embedded devices.Prince Sattam bin Abdulaziz University provided funding for this research work through project number (PSAU/2023/01/26738).FacultyReviewe
Integrating genomics into Canadian oncology nursing policy: Insights from a comparative policy analysis
Get PDFAim: To learn from two jurisdictions with mature genomics-informed nursing policy infrastructure—the United States (US) and the United Kingdom (UK)—to inform policy development for genomics-informed oncology nursing practice and education in Canada.
Design: Comparative document and policy analysis drawing on the 3i + E framework.
Methods: We drew on the principles of a rapid review and identified academic literature, grey literature and nursing policy documents through a systematic search of two databases, a website search of national genomics nursing and oncology nursing organizations in the US and UK, and recommendations from subject matter experts on an international advisory committee. A total of 94 documents informed our analysis.
Results: We found several types of policy documents guiding genomics-informed nursing practice and education in the US and UK. These included position statements, policy advocacy briefs, competencies, scope and standards of practice and education and curriculum frameworks. Examples of drivers that influenced policy development included nurses' values in aligning with evidence and meeting public expectations, strong nurse leaders, policy networks and shifting healthcare and policy landscapes.
Conclusion: Our analysis of nursing policy infrastructure in the US and UK provides a framework to guide policy recommendations to accelerate the integration of genomics into Canadian oncology nursing practice and education.
Implications for the profession: Findings can assist Canadian oncology nurses in developing nursing policy infrastructure that supports full participation in safe and equitable genomics-informed oncology nursing practice and education within an interprofessional context.
Impact: This study informs Canadian policy development for genomics-informed oncology nursing education and practice. The experiences of other countries demonstrate that change is incremental, and investment from strong advocates and collaborators can accelerate the integration of genomics into nursing. Though this research focuses on oncology nursing, it may also inform other nursing practice contexts influenced by genomics.This research was funded by the Canadian Institutes of Health Research Policy Catalyst Grant (competition number: 202210P03). The authors acknowledge the support provided by Michelle Swab, librarian at Memorial University of Newfoundland.FacultyReviewe
Dance land: Community-based dance, youth, and relationships with land
Get PDFIn 2015, the Canadian Truth and Reconciliation Commission (TRC) announced 94 Calls to Action, one of which called upon the Canada Council for the Arts to fund Indigenous and non-Indigenous artists to collaborate on projects that contribute to the reconciliation process. Reconciliation is a highly contested term, and before and after this announcement, several Canadian scholars and artists recognized Indigenous land sovereignty as central to this critical discussion. This inquiry is inspired by the work of these scholars and the spirit of the TRC Call to Action 83. In collaboration with a local ‘Native Friendship Centre’ and community youth programming, this arts-based inquiry aimed to explore how youth can use expressive movement to explore their relationships with land.
To explore these collaborative processes, I co-designed Dance Land as a method grounded in critical facilitation of community-based dance that was informed by critical place inquiry and engaged Indigenous and non-Indigenous youth participants in exploring their relationships with land. This emergent process revealed that the youth participants’ creative decision-making and my critical facilitation were rooted in embodied ways of knowing. This way of knowing is highlighted by dance scholar Barbour as thinking in movement. Thinking in movement can be attributed to relationality within Dance Land’s dance making process. Findings suggest that embodied ways of knowing can help youth explore their relationships with land. In the final chapter, implications for CYC are discussed.Graduate2025-08-1
Uncertainty and instability in social and health services impact well-being of mothers with lived and living history of substance use
No full textMothers who use substances often experience gender-based and structural inequities that can jeopardize maternal and family wellness. Instability in the availability of services, particularly during public health crises (e.g., COVID-19 pandemic), often results in changes in population health needs/funding/services, which may magnify experiences of disadvantage. Limited research has focused on times of change/crisis and its impact on maternal and family wellness. We examined the experiences of structural disadvantage, service access, and well-being among mothers who use or formerly used substances during the first year of the COVID-19 pandemic. Semi-structured interviews were conducted with 26 mothers with current or past engagement in outpatient substance use treatment programs for pregnant and parenting women in Ontario, Canada. Transcripts were analysed using reflexive thematic analysis, revealing that instability of services and decreased access to/quality of informal and formal relationships often magnified the mental and affective toll of stressors, both pre-existing and new. The impact on well-being appeared to be greater for families who were actively engaged with child protective services. Findings are discussed in relation to literature examining systemic and societal factors that perpetuate gender-based and structural inequities experienced by mothers with lived and living histories of substance use. The potential impact of changes in public health service delivery requires thoughtful and proactive attention for and by all stakeholders, including integrated attention across systems (e.g., health, social, education) that provide services to support maternal and family well-being.FacultyReviewe
Constraining Northern Cordilleran lithosphere thickness and xenolith residence times using mantle xenolith geochemistry
No full textThis study re-examines 28 mantle peridotite xenoliths and their Quaternary host lavas from near Llangorse, northwest British Columbia to determine the thickness, and reconstruct the thermal history of the Cordilleran mantle lithosphere, as well as to discover the residence time of mantle xenoliths within the host magma. Based on the equilibrium textures, the Llangorse mantle xenoliths can be separated into three distinct groups: non-sieved, weakly sieved, and strongly sieved. The sieved samples are defined by partially melted pyroxenes, with weakly sieved samples exhibiting melted rims under 50 μm wide and strongly sieved samples exhibiting broader rims. Both weakly and strongly sieved samples exhibit strong Ca zoning in olivine, with Ca concentrations increasing from core to rim, indicative of substantial heating. Closure temperatures for the xenoliths were calculated using T_Al, T_BKN, and T_REE, and varied between 829 to 941 °C, 811 to 1004 °C, and 847 to 1084 °C respectively. Depth estimates for the xenoliths obtained through Ca-in-olivine barometry (P_SC) show that only non-sieved samples and a few sieved samples yield reasonable depths between 34 and 65 (± 6) km. These depths align with both adjacent seismic measurements and the depth of equilibration of the host lava derived from SiO₂ barometry. Based on the most robust xenolith samples, I construct a model geotherm consistent with a surface heat flow of 76 (± 3) mW/m² and heat production of 1.1 (± 0.2) mW/m³. This model geotherm intersects the peridotite solidus with 300 ppm H₂O, corresponding to a lithosphere-asthenosphere boundary at 1280 (± 15) °C and 74 (± 6) km depth. Diffusion chronometry is applied to Ca zoning profiles in olivine. The results show that all the sieved samples have been heated over timescales ranging from several years to hundreds of years depending on the assumed heat source temperatures and uncertainties in the diffusion coefficient for Ca in olivine. This timescale is consistent with typical magma storage time of several years to thousands of years observed from modern volcanoes in Iceland and Hawaii. The sieved xenolith samples either have been entrained by ascending magma at an early stage, with the magma subsequently stored in the upper mantle or crustal level, or underwent heating prior to being sampled by the ascending magma.Graduat
Benchmarking Algorithms for Analysis of the Honey Bee Gut Microbiome
No full textMachine learning has emerged as a pivotal analysis technique in bioinformatics,
offering new insights that traditional statistical methods often fail to uncover. This
is particularly relevant in the study of the honey bee gut microbiome, a critical
factor in bee health and immunity, which has not yet been extensively analyzed
using these advanced computational approaches. Given the significant yet poorly
understood colony losses in recent years, studying the bee microbiome’s role
through machine learning could provide essential clues for increasing bee health and
preventing loss.
This thesis focuses on benchmarking four machine learning algorithms—random
forest, ridge regression, lasso regression, and elastic net regression—specifically for
their efficacy in analyzing the compositional changes in the honey bee gut
microbiome. These algorithms were applied to a metagenomic dataset collected
during the highbush blueberry pollination season to classify various metadata
parameters of the microbiome. Among these, the random forest algorithm
outperformed the others across several key performance metrics, including accuracy,
AUC (Area Under the Curve), kappa, and log loss, highlighting its potential as a
superior tool for microbiome analysis.
Our study further explores the challenges of machine learning in this context, such
as the risk of overfitting during hyper-parameter tuning when statistical methods
suggest minimal differences. To mitigate these issues, strategies like data
augmentation, stratified sampling, and careful partitioning of data for training and
testing were examined. These additional analyses contribute to establishing a set of
best practices for the application of machine learning in the exploration of the
honey bee gut microbiome.
Ultimately, by focusing on the benchmarking of machine learning algorithms and
delineating best practices for their application, this thesis aims to advance the
analytical techniques available for investigating the complex associations within the
honey bee gut microbiome. Such advancements are crucial for unveiling the complex
dynamics that influence bee health and for developing strategies to mitigate the
decline of bee populations critical to our agricultural systems.Graduat
A scoping review of decision-making tools to support substitute decision-makers for adults with impaired capacity
No full textBackground: Substitute decision-makers (SDMs) make decisions that honor medical, personal, and end-of-life wishes for older adults who have lost capacity, including those with dementia. However, SDMs often lack support, information, and problem-solving tools required to make decisions and can suffer with negative emotional, relationship, and financial impacts. The need for adaptable supports has been identified in prior meta-analyses. This scoping review identifies evidence-based decision-making resources/tools for SDMs, outlines domains of support, and determines resource/tool effectiveness and/or efficacy.
Methods: The scoping review used the search strategy: Population—SDMs for older adults who have lost decision-making capacity; Concept—supports, resources, tools, and interventions; Context—any context where a decision is made on behalf of an adult (>25 years). Databases included MEDLINE, Embase, CINAHL, PsycINFO, and Abstracts in Social Gerontology and SocIndex. Tools were scored by members on the research team, including patient partners, based on domains of need previously identified in prior meta-analyses.
Results: Two reviewers independently screened 5279 citations. Articles included studies that evaluated a resource/tool that helped a family/friend/caregiver SDMs outside of an ICU setting. 828 articles proceeded onto full-text screening, and 25 articles were included for data extraction. The seventeen tools identified focused on different time points/decisions in the dementia trajectory, and no single tool encompassed all the domains of caregiver decision-making needs.
Conclusion: Existing tools may not comprehensively support caregiver needs. However, combining tools into a toolkit and considering their application relevant to the caregiver's journey may start to address the gap in current supports.The students received a graduate and summer student stipend from the Brenda Strafford Foundation Chair in Geriatric Medicine, University of Calgary, and 2021 summer student scholarship from Canadian Frailty Network (CFN Project ID: SSA2021-11). This project was also funded by the Alzheimer's Society of Calgary & Pan-Canadian Palliative Care Research Collaborative, in part funded by a contribution from Health Canada, Health Care Policy and Strategies Program. The views expressed herein do not necessarily represent the views of Health Canada.FacultyReviewe