Innovative Projects Realized

Workplace Mental Health Promotion in Small Businesses: A Scoping Review and Feasibility Study

Workplaces have an important role to play when it comes to promoting the mental health of employees. Whereas we have an extensive body of evidence about how to develop and deliver interventions to promote employee
wellbeing and mental health in large organizations, leaders in organizations often face steep barriers when adapting those interventions in small businesses. This research will fill gaps in workplace mental health literature
through two complementary research phases. Phase 1 involves a rapid, systematic review of academic studies that describe mental health interventions that were delivered within small workplaces. Phase 2 involves semiinterviews
with stakeholders who have experience implementing interventions within real-life small businesses. Findings will offer practical strategies for designing, implementing, and evaluating programs focused on employee
mental health within small businesses. Knowledge mobilization efforts, including written reports, presentations, and publications, are planned to share our research across various domains.

Exploration de nouveaux systèmes salins pour la purification du graphite

La purification du graphite vers des formes à haute pureté implique la transformation des minéraux de gangue afin de pouvoir séparer les éléments le constituant du graphite. L’une des voies considérée pour cette transformation implique l’utilisation de sels capables de réagir notamment avec les silicates. L’objet de ce stage est l’utilisation de mélanges de sels pour un tel traitement. Cette approche pourrait notamment d’aller vers des conditions opératoires moins énergivores, notamment par l’utilisation de températures plus faibles. Différents essais à l’échelle laboratoire sont prévus afin d’explorer ces systèmes salins, leur réactivité et leurs propriétés.

Paper electrochemical devices for bacterial analysis

Background- Bacterial analysis in the field is complex and often imprecise. It is therefore difficult to detect and identify a specific strain in a timely manner, which complicates the choice of an optimal treatment regimen. This is a critical issue and bacterial infection are taking an increasing toll on the population and the health services. Having access to a widely disseminated, easy-to-use and user-friendly device would be a clear asset in this task.

Objective- The purpose of the project is to design a paper-based device for the sampling and detection of bacteria in unprepared samples. The motivation for using paper as a chip substrate are its low cost, its high availability and its capillary action that can be used to sample liquids passively. The bacterial detection will be completed using electrochemical methods with nanofunctionalized electrodes. Important aspects of the device will be i- passive sampling, ii- on board detection and iii- encapsulation of the sample in a hydrophobic layer that will limit contamination and drying of the sample.

Investigation of the Susceptibility of High Voltage Hydroelectricity Transmission Lines to Wildfire Smoke

The generation of clean hydroelectric power occurs often in locations far from consumers. High voltage transmission lines that traverse vast terrains are employed for the transmission of the generated clean electric energy. They are exposed to various environmental conditions, including wildfire smoke. Wildfire smoke has the potential to decrease the breakdown strength of air insulation, which serves as the primary insulation system for overhead power lines, leading to outages. Investigating the effect of smoke on the insulation strength of air is crucial to increase the reliability of the power transmission network and reducing wildfires ignited by electrical discharges. The proposed research studies the discharge characteristics and insulating properties of air under different smoke conditions. Laboratory experiments will be designed and conducted to determine the dielectric characteristics of air under various smoke conditions. In addition, guidelines for safe live line maintenance and mitigation techniques will also be established.

Optimizing Credibility Complements in Property and Casualty Insurance: A Data-Driven Approach for Enhanced Premium Estimation

Within the field of property and casualty insurance (P&C), one of the most pertinent dilemmas is the estimation of the insurance premiums required to accommodate potential losses. Due to the inherently random nature of losses, one of the tools utilized within P&C insurance is credibility theory, which utilizes a weighted average between past portfolio data and the portfolio’s general risk class. This allows insurance companies to give greater weight to more reliable data and theoretically increase the accuracy of their premiums.
However, the numerous options for a portfolio’s risk class, also denoted as the complement of credibility, create a subjective pricing model that may not be optimal for a given portfolio. Currently, the complements of credibility that are employed are finite and limited to traditional criteria. Therefore, generating a model capable of optimizing the complement of credibility may be beneficial by using more data from separate risk classes. Gathering data from various business units and portfolios could generate a prediction model that calculates a custom complement for each portfolio. This would increase the general relevancy of the complement of credibility and the associated profitability while ensuring that insurance premiums are more accurate and reliable.

Innovative energy efficiency in residential buildings: main blower system for interroom air transfer

Residential buildings contribute approximately 17% to Canada’s overall primary energy consumption, with the majority allocated to space heating, particularly space heating. The proposed research will focus on optimizing
energy efficiency through inter-room air transfer, utilizing a central blower system. SmartCocoon’s product includes fans at each room diffuser (AKA vent/register) such that heat can be strategically reallocated to occupied
spaces. While past work focused on modelling potential energy savings and comfort improvement, this project involves detailed field testing of various control strategies in an occupied home. The outcomes will guide
recommendations to our partner, Smart Cocoon, for implementing energy-efficient control strategies in their interroom air movement system. By considering factors such as room occupancy, solar heat gain, and existing HVAC
configurations, the system will achieve significant reductions in operational costs while optimizing comfort and energy conservation, aligning with the broader goal of sustainable residential energy consumption.

Visual SLAM for Navigation & Mapping

Autonomous vehicles (AVs) promise to enhance safety, reduce emissions, and improve transportation system efficiency and reliability. The growing demand for AVs is shaping the future of the automotive industry by transforming the in-vehicle experience and paving the way for large-scale implementation of autonomous driving. AV technology requires onboard intelligence relying on sensors and systems such as global navigation satellite systems (GNSS), including GPS, vehicle motion sensors and remote sensing systems, including cameras, light detection and ranging (LiDAR) and radar. AVs capable of sensing the environment and navigating without human input require robust, high-precision positioning at the decimeter level of accuracy under all operational environments. During this internship, we will devise a vision-based navigation (VBN) module enabling decimeter level of positioning accuracy relying on centralized visual-inertial odometry supported by HD-Maps aiding and a deep Learning-based outlier rejection to mitigate the effect of dynamic objects. This project will advance visual self-localization and mapping for AVs and will contribute to the autonomous systems research at Queen’s. It will also elevate the intern’s expertise in areas of growing demand.

Investigating the Effects of Emotion on Visuo-Somatosensory Integration by Using an AI-Powered Markerless Motion Capture System

We use all forms of sensory information when directing our movements to ensure that we reach our target. Conversely, when we have conflicting tactile and visual streams of information, like shaving in the mirror, the brain tends to inhibit the tactile information in order to favour our visual information. However the extent to which this is observed, based on our emotional response to the tactile sensation, is not understood. We plan to build an AI-powered markerless motion capture model (MMC) to track a goal-directed movement across pleasant and unpleasant surfaces, while analyzing electroencephalography (EEG) to investigate how the brain responds. By bringing our AI experience to the host institution, we hope to leave behind accessibility to motion capture technology through a pipeline and AI-powered MMC model that they are able to use for further research. The host institution has also offered a course as well as training with utilizing EEG technology, which will be utilized in conjunction with the results of this project, for further research projects and an undergraduate thesis at the home institution.

Neural Pathways to Health: Deep Learning Applications in Medical Science

Machine Learning algorithms proved to be very helpful in medical science in the presence of enormous amounts of data. Particularly, it analyzes medical data, such as images, genetic information, and patient records, aiding healthcare professionals in faster and more accurate diagnoses. Deep learning has gained significance in automated report generation, expediting the diagnostic process and improving overall accuracy for more effective treatment plans. Specifically, applications like interpreting chest x-ray images highlight the need for precision in the absence of practitioners. To ensure trustworthiness and reliability, incorporating explainability concepts is crucial.
Machine Learning customizes medical interventions based on individual patient characteristics, considering factors like genetics, lifestyle, and health history. This personalized approach enhances treatment effectiveness, minimizes side effects, and boosts patient satisfaction. In the realm of dermatopathology research, interpretable deep learning techniques are applied to analyze histological images of common skin cancers such as intraepidermal carcinoma (IEC), squamous cell carcinoma (SCC), and basal cell carcinoma (BCC). The study involves classifying skin tissue into 12 dermatological classes, including structures like sweat glands and hair follicles, showcasing the potential of automatic machine analysis in this field.

Investigation of Consumers’ Perception Toward Sustainable Product Attributes of Consumer Electronics: A Canadian Case Study

The proposed research attempts to understand Canadian consumers’ perception toward sustainable product attributes of consumer electronics products and factors affecting the consumers’ choices concerning sustainable consumer electronics during product purchase. The study uses the Theory of Planned Behavior (TPB) as its theoretical framework and further attempts to extend the TPB by incorporating additional constructs such as environmental concern and environmental knowledge in it. Data will be collected using a mixed method consisting of a self-administered questionnaire survey and focus groups. The data collected from the survey and focus groups will be analyzed with the help of structural equation modeling (SEM) and the content analysis technique, respectively.

Using Machine Learning Techniques to ImproveAutomatic Keyword Extraction from Textual Web Content

This internship aims to implement an advanced system for automatic keyword extraction from textual web contents. Keyword extraction not only provides a concise and salient representation of a document, but also can be used in web-based applications such as efficient indexing, intelligent tag recommendation, and contextual advertising. In this research project, a hybrid keyword extraction approach is proposed which combines the knowledge learned from an annotated dataset of manually extracted keywords with the knowledge
acquired by processing a huge unannotated corpus of data, using machine learning techniques.

Improving Handover Management in 5G New Radio

The demand for higher bandwidth and ultra-lower latency has increased due to the rise of wireless devices. To meet these challenges, fifth generation (5G) new radio (NR) offers enhanced bandwidths with higher data rates and ultra-lower latency for real-time applications. The ultra-dense networks in 5G NR can improve the coverage. However, it can cause problem of frequent user handovers in inter- or intra-network, leading to high delays and degraded quality of service (QoS). The handover in mobility is the primary key performance indicator (KPI), which requires reliability and lower latency. Inefficient network deployment can lead to issues in the user handover process, especially in high mobility. Therefore, optimizing the handover in 5G NR for high-mobility users is crucial. The research project will set the ground for long-term partnerships between Lakehead University and MUET.