Innovative Projects Realized

Metabolomic Characterization of Wild Canadian Haskap (Lonicera caerulea var. villosa) Germplasm

Haskap is an emerging berry crop in North America known for its unique flavour and high nutritional value. Commercial haskap cultivars have received considerable attention from researchers, however, Canada’s native wild haskap populations remain largely unstudied. This project aims to comprehensively analyze the metabolomic profiles of 150 wild Canadian haskap accessions maintained by the Fruit Breeding Program at the University of Saskatchewan. Leaf samples collected from the Canadian germplasm, as well as some advanced selections and commercial cultivars, will be analyzed at the University of Ljubljana using their established haskap metabolomic methods to identify and quantify metabolites. The resulting metabolomic profiles will be used in conjunction with existing phenotypic data to identify accessions with high potential to improve the nutritional profiles of future haskap cultivars as well as to identify biomarkers that can be used to accelerate haskap breeding at the University of Saskatchewan. This collaboration will benefit both institutions by combining the unique genetic resources of the University of Saskatchewan with the specialized metabolomics expertise of the University of Ljubljana, accelerating breeding efforts, fostering future joint research initiatives, and enhancing the training and scientific capacity at both universities.

Web-based Goal Management Training™ in older adults with cognitive impairment: a randomized-controlled trial examining feasibility and efficacy – Year two

As the Canadian population is aging, more and more Canadians will show cognitive decline. Aging and certain types of
neurological disorders is often associated with deficits in executive functions: goal maintenance, planning, task
switching and attention. These functions are critical for the maintenance of functional independence. Few validated
rehabilitation approaches for these types of deficits exist. One rehabilitation approach, Goal Management Training™
(GMT), has shown promise. In its standard implementation, GMT is led in small groups. Although this approach is
effective, it has significant practical limitations (limited accessibility and high cost). The goal of the proposed research
is to design an automated, web-based GMT training program that can be delivered remotely, to assess its feasibility in
older adults and patients with deficits in executive functions and to compare its efficacy to an active control group
that would participate in an online activity that we do not expect to improve cognition.

Health and Illness in the works of Friedrich Nietzsche and Georges Canguilhem

My master’s thesis researches the fundamental and original concepts of health and illness in the works of Friedrich Nietzsche and Georges Canguilhem, as well as the reading of Nietzsche by Canguilhem which informed his writings. I plan to study for one year (Oct. 2025 – Aug. 2026) at the Bauhaus-Universität Weimar, Germany, under the supervision of Professor Henning Schmidgen. He will guide my research on both thinkers, for whom Weimar and Paris are essential research locations. The presence of the Nietzsche Archives in Weimar is a major asset. This immersion will be my first university experience abroad, conducted entirely in German. I aim for a C1 language level and the acquisition of specific philosophical vocabulary, while also broadening my worldview and network. This stay could also establish a lasting bilateral intellectual relationship between Université Laval and Weimar. My plan includes validating my thesis topic, targeted archive consultation, and an intensive German course in Berlin, followed by my move to Weimar for courses and research. I’ve scheduled trips to the Canguilhem Archives in Paris and regular visits to the Nietzsche Archives. The university’s multidisciplinary approach will broaden my horizons. My goal is to finalize my thesis by August 2026.

ML-Enhanced SOAR Framework for Proactive Threat Response in Managed Security Operations

Security Operations Centers today grapple with overwhelming alert volumes, fragmented toolchains, and manual response processes that impede timely threat containment. Analysts must pivot between multiple SIEM and EDR consoles, manually enrich indicators, and open tickets one by one, introducing delays that adversaries exploit to dwell undetected. Moreover, static severity tags lack the nuance to prioritize truly critical events, while developing and maintaining effective response playbooks is laborious and error-prone.
This project, in collaboration with GlassHouse Systems and the University of Guelph, will deliver:
• A unified SOAR integration layer that normalizes alerts and actions across all client SIEM, EDR, threat-intelligence, and ticketing systems;
• An ML-powered risk-scoring service trained on historical incident outcomes and enriched threat data to assign every alert a dynamic priority score;
• Automated response workflows that invoke the ML scores to escalate high-risk threats, retire low-risk noise, and guide analysts through ambiguous cases with complete context.
By embedding machine learning at the core of playbook orchestration, this research will accelerate mean-time-to-detect and mean-time-to-respond, reduce false-positive workloads, and establish a reproducible, metrics

Assessing the Impact of B.C.’s PST on Industrial Investment: A Policy Analysis

This project aims to examine the role of British Columbia’s Provincial Sales Tax (PST) in shaping capital investment decisions within the industrial and advanced manufacturing sectors. Specifically, it will explore how the taxation of machinery and equipment under the PST has evolved over the past 20 years (2004–2024), and how the current tax structure may serve as a barrier to productivity by enhancing investments.
Through a combination of policy review, data analysis, and interprovincial comparison, the project will assess the extent to which B.C.’s PST framework may be discouraging industrial reinvestment. The findings will inform the development of targeted, evidence-based policy recommendations, particularly around potential PST exemptions or reforms, that could incentivize capital investment and improve economic competitiveness.
The final deliverable will be a policy-oriented report for the Greater Vancouver Board of Trade (GVBOT), supporting its Agenda for Growth by offering actionable insights to guide provincial advocacy efforts on tax reform and industrial policy.

Codesigning a stepped care model of mental health supports for Canadian Veterans and their families

This project will engage Veterans, Families and service organisations across Canada to co-design a continuum of care that supports the mental health and substance use health of Veterans and Families. A series of 3-4 engagement sessions will be held with approximately 50 Veterans and families, as well as a 2-3 workshops series with approximately 30 service providers, system leaders and advocates. The process is designed to identify key mental health and substance use services currently available to Veterans and families, as well as opportunities and service needs/gaps. The process will engage key interest holders to co-design recommendations to increase access, flexibility, collaboration and ease of navigation across programs and services. Key findings will shared wtih participating organizations and with the Veteran and family communities.

Bridging the Gaps: Identifying Practical and Emotional Needs in Cancer Care

Look Good Feel Better (LGFB) is a national charitable organization dedicated to enhancing the quality of life for people facing cancer. We do this by educating and empowering individuals to manage the appearance-related and psychosocial impacts of cancer and its treatment. Our workshops and resources are designed to help people feel more like themselves again by offering practical, compassionate support for dealing with visible and often distressing side effects—such as hair loss, changes to the skin, and other physical impacts. LGFB programs are offered free of charge, both online and in person, at over 80 hospitals and cancer centres across Canada.

Our programming is grounded in research that shows how psychosocial interventions, like LGFB’s workshops, can reduce depression, improve self-esteem, and enhance quality of life for people with cancer. While our work has focused primarily on the visible side effects of treatment, LGFB also aims to fill broader gaps in supportive care that are not always addressed within the healthcare system.

Despite growing evidence of the benefits of psychosocial support, access remains inequitable and fragmented across Canada. People facing cancer often lack access to low-intensity, universal supports that can prevent more serious mental health issues from emerging. Many report receiving information about physical side effects of treatment, but fewer feel they have adequate support for emotional or practical concerns. This unmet need extends well beyond the active treatment phase into survivorship.

LGFB is seeking to understand which types of supportive care are most needed and valued, and where the biggest gaps lie across the cancer journey. This research represents an innovative shift for LGFB: from solely delivering supportive programs to also contributing to national knowledge about how to best design and deliver psychosocial care for people affected by cancer.

Web-based Goal Management Training™ in older adults with cognitive impairment: a randomized-controlled trial examining feasibility and efficacy

As the Canadian population is aging, more and more Canadians will show cognitive decline. Aging and certain types of
neurological disorders is often associated with deficits in executive functions: goal maintenance, planning, task
switching and attention. These functions are critical for the maintenance of functional independence. Few validated
rehabilitation approaches for these types of deficits exist. One rehabilitation approach, Goal Management Training™
(GMT), has shown promise. In its standard implementation, GMT is led in small groups. Although this approach is
effective, it has significant practical limitations (limited accessibility and high cost). The goal of the proposed research
is to design an automated, web-based GMT training program that can be delivered remotely, to assess its feasibility in
older adults and patients with deficits in executive functions and to compare its efficacy to an active control group
that would participate in an online activity that we do not expect to improve cognition.

Real-Time Machine Learning Integration with MES for Waste Reduction, Quality Control, and Production Optimization in Flexible Packaging at Transcontinental

This project focuses on the real-time integration of machine learning (ML) models with the Manufacturing Execution System (MES) at Transcontinental (TC) to enhance quality control, reduce production waste, and optimize operations in flexible packaging manufacturing. TC’s current MES infrastructure ensures traceability and workflow coordination but lacks predictive analytics capabilities, leading to a reactive approach where quality issues and process inefficiencies are often addressed only after causing significant material loss, downtime, and customer dissatisfaction.

To address this gap, the project introduces a strategic initiative aimed at transforming live production data into actionable insights through ML-driven predictive modeling. The solution will enable early detection of process anomalies, suggest real-time corrective actions, and evolve continuously with new data patterns. Historical and live data from critical processes such as extrusion, printing, and lamination will be leveraged to train, validate, and deploy ML models directly into the MES environment, enhancing decision-making at the operational level. By embedding intelligence into MES workflows, the project aims to minimize scrap rates, improve first-pass yield, reduce energy usage, and increase overall equipment effectiveness (OEE).

Expected outcomes include significant reductions in waste and energy consumption, enhanced process stability, and improved responsiveness on the shop floor. This initiative positions TC at the forefront of Industry 4.0 by integrating AI into core manufacturing operations, supporting long-term innovation, sustainability, and digital transformation in the flexible packaging sector.

Medventions Atlantic Fall 2025 – General Surgery

The Medventions Atlantic program based on the proven model from Sunnybrook Health Sciences Centre enables high-performing students and recent graduates from diverse academic backgrounds to engage directly with frontline clinical and innovation challenges. The interns are embedded within clinical environments to identify unmet needs, apply early-stage innovation thinking, and co-develop solutions with clinicians, patients, and administrators.

Génération de données synthétique en utilisant le concept de confidentialité différentielle et mise en lien avec d’autres mesures de confidentialité

Beneva est une entreprise québécoise à caractère mutualiste, qui offre des produits en assurances et en services financiers. Dans le contexte de la loi 25 adoptée par le gouvernement du Québec pour renforcer la protection des renseignements personnels, en adaptant les règles en matière de confidentialité aux nouvelles réalités numériques, Beneva a été soumise à des obligations accrues en termes de gestion, de collecte et de divulgation des données personnelles. L’une des solutions envisageables pour se conformer à ces exigences est l’anonymisation des renseignements personnels (RP). Cette méthode permet de transformer les données de manière qu’elles ne puissent plus être associées à une personne identifiable, tout en préservant leur utilité pour l’analyse statistique et la prise de décisions.
Afin de renforcer ses processus de protection des données confidentielles, Beneva souhaite tirer parti d’un partenariat avec Prof. Anne-Sophie Charest du département de mathématiques et de statistique de l’Université de Laval, en s’alliant pour approfondir les méthodes d’anonymisation des données. Beneva utilise déjà l’approche de la génération de données synthétiques, basée sur des modèles génératifs CTGANs [1] et TVAE [2]. Cependant, plus le degré de conservation des propriétés statistiques est élevé, plus le risque d’identification est élevé, même avec des données synthétiques. Un juste équilibre est alors souhaitable. Beneva vise ainsi à améliorer les méthodologies existantes en introduisant dans le processus de génération de données synthétiques un paramètre qui permet de générer des données avec le degré de confidentialité/utilité souhaité selon le besoin de partage. Plusieurs méthodes sont à tester, dont l’ajout de la notion de confidentialité différentielle, lors de la génération des données synthétiques, avec l’objectif de satisfaire certaines métriques de divulgation existantes qui sont calculées a posteriori.
En plus de se conformer aux réglementations sur la protection des renseignements personnels, le projet représente des avantages économiques, tels que :
1. Amélioration des performances des modèles d’IA.
2. Facilitation du partage et de la collaboration : qui permet d’Accélérer des partenariats et collaborations internes et externes.
3. Réduction des délais de mise en marché : en évitant les contraintes liées à l’accès aux données réelles.
Le projet a aussi des avantages sociaux, tels que :
1. Renforcement de la protection des données des clients.
2. Contribution à l’innovation et à la recherche : en permettant aux chercheurs et aux équipes internes de tester de nouvelles approches sans compromettre la sécurité des données.
3. Accessibilité et démocratisation des analyses de données.
4. Soutien à la mission mutualiste.

Advances in sensor fusion and gap-filling for satellite based analysis ready surface reflectance data

This project aims to develop a physics-guided deep learning framework that can be used to further enhance the robustness of surface reflectance forecasting in Planet Fusion. We want to improve and refine existing temporal-driven gap-filling techniques in handling extensive cloud cover and dynamic land changes to avoid delays and reduced reliability of delivered insights to the extent possible. The integration of domain knowledge with advanced AI techniques will allow Planet to deliver more robust, continuous, and predictive Earth observation products. This addresses a core need for the company and its clients in agriculture, climate resilience, and land management, who increasingly demand reliable uninterrupted near real-time surface reflectance solutions.