Projets novateurs réalisés

Cartographier pour préserver : l’IA au service des milieux humides du Québec

Les milieux humides assurent des fonctions écologiques essentielles : préservation de la biodiversité, séquestration du carbone, régulation des inondations et purification de l’eau. Pourtant, ces écosystèmes sont en déclin rapide, disparaissant à un rythme trois fois plus élevé que les forêts, sous la pression de l’urbanisation, de l’agriculture intensive et de l’extraction des ressources. Bien que de vastes bases de données issues de la télédétection soient disponibles pour leur cartographie, les méthodes traditionnelles de traitement restent limitées, souvent inefficaces pour répondre aux besoins des gestionnaires et des décideurs. C’est dans ce contexte que s’inscrit le projet de recherche mené en partenariat entre l’équipe du professeur Mostafavi du Centre de recherche en données et intelligence géospatiale (CRDIG) et Canards Illimités Canada. Ce projet vise à explorer et à intégrer des approches avancées d’intelligence artificielle (IA) — notamment l’apprentissage profond — pour améliorer l’identification, la délimitation, la classification et la cartographie des milieux humides. L’objectif est de générer des cartes précises, à jour et adaptées aux spécificités écologiques du Québec, afin de renforcer les efforts de conservation et de planification territoriale du partenaire et, plus largement, de soutenir une gestion durable de ces écosystèmes vitaux.

The effect of an automated horse feeder on horse behaviour and welfare

Free-ranging horses forage for 16-20 h/d, travel up to 20 miles and interact with conspecifics as they choose. Conversely, domesticated horses are often kept in individual stalls, meal-fed, turned out for short durations and not given a choice with whom they interact. These living conditions challenge efforts to mimic horses’ natural feeding rhythms and balance nutritional intake. One solution is to provide ad libitum hay to horses, but this may lead to obesity, prevalent in westernized societies, and increased risk of metabolic diseases. Weight management strategies involve keeping horses in confined environments with reduced access to forage. To address this challenge, automated feeding devices allow frequent feeding mirroring natural patterns while reducing caloric intake and the labour required to care for horses.
The Nordic AutoFeeder© (NAF), is designed to reduce labour on farm while providing frequent feedings throughout the day. This study will investigate how social behaviours among horses may be affected by the NAF and if the distance walked each day differs compared to ad libitum feeding. These evidence-based outcomes will enable them to provide recommendations to their customers on feeding time budgets and promote their product with confidence.

Développement de modèles d’apprentissage pour la prédiction des routes de livraison conviviales pour les livreurs via l’intégration de leurs connaissances : étude de cas à Montréal

Ce projet vise à améliorer la qualité des itinéraires de livraison proposés aux livreurs dans les villes, en tenant compte de leurs préférences, habitudes et connaissances du terrain. Grâce à l’intelligence artificielle et à l’analyse de données réelles collectées à Montréal, la stagiaire développera des modèles capables de prédire des tournées de livraison plus adaptées à la réalité, faciles à suivre et moins sujettes aux modifications humaines. Ce travail permettra d’optimiser les opérations logistiques, de réduire les coûts et les émissions polluantes, tout en rendant le travail des livreurs plus fluide et agréable. Le projet contribuera ainsi à une logistique urbaine plus intelligente, durable et centrée sur l’humain.

Advanced Optimal Production Scheduling for Salt Manufacturing at Compass Minerals

This project develops a decision-support tool to improve production scheduling at Compass Minerals’ salt plant in Goderich, Ontario. The plant makes multiple grades of salt on connected production lines, facing changing demand and tight operational limits. Currently, schedules are created manually or with simple rules, which limits flexibility and efficiency. We aim to build an optimization model that generates efficient, realistic schedules while accounting for constraints like demand, equipment capacity, changeover times, and inventory. To handle uncertainty, we will use stochastic programming techniques. The final tool will be tested with plant data and refined with input from plant staff. It is expected to improve scheduling, reduce costs, and help the plant respond better to changes in demand, supporting both profitability and sustainability.

Modulating chromatin remodeling to reverse immune evasion in ovarian cancer

Immunotherapy has achieved remarkable responses in cancers that were once fatal. However, Ovarian Cancer (OC) clinical trials have shown disappointing responses with immune checkpoint blockade (ICB). Our research team is studying how to target the factors influencing interactions between OC and immune cells to improve the treatment of OC. One particular target is the SWI/SNF complex, involved in modifying the structure of DNA. Previously, we discovered that inhibiting this complex increases the ability of OC cells to be recognized by the immune system. Another protein target, called DNA methyltransferase (DNMT), has also been shown, when inhibited, to increase the expression of immune-related genes. Based on these findings, targeting both factors is a logical therapeutic strategy to sensitize OC to immunotherapy for a more effective response. Therefore, we propose to combine DNMT and SWI/SNF inhibitors with ICB. The JGH Foundation is a non-profit organization committed to advancing research and advocacy in OC. Through this collaboration, the Foundation seeks to address the urgent need for innovative treatment strategies, particularly for women with chemoresistant disease. This proposal offers the potential for the JGH to be leaders in developing novel immune-based treatment approaches to improve the lives of women affected by OC.

Combining Music and Non-Invasive Brain Stimulation to Improve Gait in Parkinson’s Disease

Parkinson’s disease (PD) is a progressive neurological condition that affects movement, often leading to slower walking, difficulty maintaining rhythm, and increased risk of falls. This project investigates whether combining two approaches—transcranial direct current stimulation (tDCS), a form of non-invasive brain stimulation, and rhythmic auditory stimulation using music—can help improve walking in individuals with PD. The study is divided into two phases. In the first phase, we will examine the immediate effects of a single session of tDCS over the supplementary motor area (a brain region involved in planning movement) combined with beat-based music while walking. In the second phase, we will explore whether multiple sessions of this combined stimulation lead to longer-lasting improvements in gait. By comparing outcomes from individuals with PD and healthy older adults, we aim to better understand the specific benefits of this intervention. The results may support the development of accessible, low-cost strategies to improve mobility and quality of life in people living with Parkinson’s disease, while also enhancing our understanding of how rhythmic and brain-based interventions influence motor control.

Population Viability Analysis of Blanding’s Turtles in Nova Scotia

Freshwater turtles are one of the most at-risk vertebrate groups, with nearly two-thirds of species facing extinction risk. Nest predation by mammals is a major threat to freshwater turtles, which has resulted in community-based nest protection efforts in many regions around the world, including Nova Scotia, Canada. Although widely implemented, the efficacy of nest protection programs have been questioned, especially for long-lived species such as freshwater turtles. While many agree that protecting nests improves survival of eggs and hatchling life stages, and thereby increasing recruitment into the population, it is not yet clear whether these individuals reach adulthood to reproduce and reduce chronic rates of population decline. As a result, conservation organizations face a critical knowledge gap when prioritizing recovery actions. We aim to develop an updated population viability analysis (PVA) to evaluate how nest protection efforts improve long-term viability of the endangered Blanding’s Turtle (Emydoidea blandingii) in Nova Scotia, Canada using a unique 20-year community-based dataset. Such assessments are necessary for guiding future community engagement and managing species of conservation concern, both within canada and globally.

AC Solar photovoltaic module testing for residential applications

The project’s purpose is to integrate Heliene’s solar photovoltaic (PV) panels with power electronics to create an integrated, high-performance solar module product for residential applications. The resulting AC solar module removes the need for external electronic components such as DC optimizers, inverters, diodes, and rapid shutdown devices. By integrating the external components, overall system reliability is improved, as well as reducing installation time and maintenance costs. Equally important, the integrated solar module harvests more solar energy under typical lighting conditions and challenging light situations, such as panel soiling and heavy shading. Prototype AC modules, along with the conventional modules, need to be installed in a field for a pilot test run to determine the impact of partial shading on both systems. This project will provide that basic pilot testing. Heliene can leverage this data to commercialize their new innovation.

Implémentation de la planification stratégique pour le Réseau des Grandes Cultures du Québec (RGCQ)

Implémentation de la planification stratégique pour le Réseau des Grandes Cultures du Québec (RGCQ)

Cannabidiol (CBD) as a neuroprotectant in Parkinson’s disease (PD) through the Nrf2 pathway

Parkinson’s disease (PD) affects movement, much like an orchestra losing its conductor; signals become uncoordinated, leading to tremors, stiffness, and difficulty moving. This happens when dopamine-producing cells in the brain’s substantia nigra die. A major cause of this cell death is oxidative stress, an imbalance between harmful molecules called free radicals and the body’s protective antioxidants. One of the body’s main defenders against oxidative stress is a protein called Nrf2, which activates antioxidant enzymes to protect brain cells. However, under prolonged stress, Nrf2 itself can become damaged and lose its protective function.
Cannabidiol (CBD), a non-psychoactive compound from Cannabis sativa, is a strong antioxidant that can cross into the brain. It may help restore the function of Nrf2 and shield brain cells from damage. In this study, I will examine how Nrf2 behaves in PD patient brain tissue and explore how CBD may restore Nrf2 in both human brain samples and an experimental mouse model of PD. By restoring this antioxidant defense system, CBD could potentially slow or prevent further brain cell loss. This research may lead to new, targeted therapies using CBD to address one of the root causes of PD progression.

Processus d’implantation d’AToM – prolongement

Le Centre d’archives régional de Portneuf (CARP) a pour mission de préserver et de faire connaître les archives liées à la région. En 2024, il a franchi une étape importante en se dotant d’une plateforme numérique (appelée AToM) pour mieux gérer, protéger et partager ses collections.

Ce nouvel outil a nécessité une adaptation importante pour l’équipe du CARP. Grâce au soutien de Mitacs, un stagiaire a pu développer une procédure claire pour accompagner cette transition. Cela a permis de mieux organiser les archives numériques, de moderniser les méthodes de travail et d’assurer un traitement plus efficace des documents.

Même si le transfert complet des archives vers AToM est toujours en cours, le travail réalisé a déjà permis d’avancer rapidement. Le CARP prévoit que toutes ses données seront disponibles sur la plateforme d’ici l’été 2026. En attendant, AToM est déjà accessible à l’interne pour les employés, bénévoles et chercheurs.

Des tests ont révélé certains défis à venir, notamment pour rendre la plateforme plus conviviale pour le grand public, qui pourrait être dérouté par le langage spécialisé des archivistes. Le CARP devra aussi réfléchir avec soin aux enjeux juridiques et éthiques liés à la mise en ligne de documents sensibles, comme des photos ou des vidéos.

Pour poursuivre ce virage numérique de façon responsable, le CARP souhaite s’entourer d’expertise en archivistique et en diffusion d’archives, afin d’assurer une représentation juste, respectueuse et accessible du patrimoine régional.

Quantum Machine learning for medical imaging – radiology, ophthalmology and diabetes prediction

This project explores how quantum computing and artificial intelligence (AI) can work together to improve how we detect diseases using medical images. Medical imaging, like X-rays, clinical photos, and eye scans, are essential for diagnosing many health conditions. However, analyzing these images quickly and accurately remains a challenge, especially as the amount of data continues to grow.

By combining the power of AI with the emerging capabilities of quantum computing, this research aims to develop Quantum Machine Learning algorithms in order to create smarter, faster, and more reliable tools for identifying diseases. The project focuses on three key areas: detecting lung conditions from chest X-rays, identifying diabetic foot ulcers from clinical images, and spotting retinal diseases from eye scans.

Ultimately, this project supports the development of next-generation healthcare tools that could lead to earlier diagnoses, better treatment decisions, and improved patient outcomes. It also helps build Canada’s leadership in digital health and quantum innovation by training future experts in these fast-growing fields.