Innovative Projects Realized

Stratigraphic Architecture of a Carbon Sequestration Target: Aquistore Site, Saskatchewan

This project will train a Master’s geology student to use a variety of geological and geophysical datasets and concepts to define the stratigraphic architecture of deeply buried sandstones used for CO2 sequestration at the Aquistore site in Saskatchewan. The extent to which these sandstones form a single continuous body, or are compartmentalized, will affect their ability to transmit and store the injected gas. These characteristics were defined nearly 500 million years ago when these strata were deposited. Results will inform ongoing monitoring and modeling efforts at Aquistore whereas the methods will be transferable to other proposed CO2 sequestration sites in Canada and abroad.

Generating Synthetic Bank Transaction Sequences with GAN-based Models

Financial institutions gather vast amounts of data from our transactions. However, using this data directly can risk our privacy. A potential solution is generating “synthetic data”, which contains real data information without copying the original data. Generating this data, especially for bank transaction sequences, is challenging. During our research internship, we will use recent developments in deep generative models to design a generative model that can create high-quality bank transaction sequences. This research project opens doors for Verafin to team up with outside experts and lets them show potential clients how their product works using realistic, risk-free examples.

A proposed hybrid machine learning model for fault-type classification and detection in a 10 MW wind farm based on synchrophasor and weather data

Incorporating renewable energy sources, such as wind and solar farms, have posed a challenge to electrical engineers due to their intermittent nature as the time and amount of energy generated depends on the weather. For this reason, grid stability and predicting the behaviour of renewables has been increasingly more important. In this study, we propose to train a machine learning algorithm on weather data, synchrophasor readings (voltage, current, and phase), and turbine monitoring data from a 10 MW wind farm located in PEI to detect and classify faults in their wind turbines. We will use the algorithm to develop a protection plan for their wind turbines based on these results.

Targeting TRAF1 in psoriatic arthritis

There is no cure for psoriatic arthritis (PsA) and current treatments aim at reducing inflammation, slowing the progression of the disease, and avoiding severe damage to the skin and joints. Immune cells contribute to inflammatory cascades in the joints and the skin that produce inflammatory cytokines and lead to severe tissue damage. Current therapies are geared towards modulating the patient’s immune response or blocking the action of specific cytokines; unfortunately, many patients do not respond to treatment or suffer from significant side effects. This necessitates the development of new therapies that would specifically target the immune components that are involved in PsA pathogenesis and “pump the brakes” on their ability to produce the damaging cytokines.

TRAF1 is a protein linked with increased risk for rheumatoid and psoriatic arthritis. Dr. Abdul-Sater’s lab has recently discovered that this protein is important for cytokine production in immune cells. Dr. Abdul-Sater’s team created a modified version of this protein that improved its ability to limit inflammation and block cytokine production. In this proposal, and in collaboration with the Krembil Foundation, Dr. Abdul-Sater will test whether modifying this protein the same way in PsA patient cells will reduce their ability to produce damaging cytokines…

Market and Technology Roadmap Validation Framework for Software Services

Founding a technology startup company is never an easy endeavor. There are major challenges on both the technology side and the business/commercialization side. For Tutela Technologies Ltd. it appears that the commercial validation of the technologies invented is proving to be the largest challenge. The initial internship and research will focus directly on Tutela Technologies Ltd. Tutela’s case will be used to create a framework and roadmap for establishing commercializing success and focusing R&D activities. In the second internship, using the newly created framework, these validations will vary among the companies in terms of markets, specific products and technologies that in progress or already developed. The main problem that will be addressed in this internship is identifying market opportunities and strategies for commercially validating the technologies that the Tutela Technologies and some of the other Wesley Clover portfolio companies have developed. The benefit to Tutela and Wesley Clover will be the commercial success of the products and services created. There is a need to create a framework that can identify roadmaps to commercialization in order to have a financially sustainable future and successfully grow.

Fanny Lo Internship

Project 1
Circle Cardiovascular Imaging is a software medical device Manufacturer based in Calgary Alberta, Canada. Circle Cardiovascular
imaging creates multiple products for the analysis of patient images for assisting the diagnosis of cardiac and neurological
diseases. cvi42 Cardiovascular Magnetic Resonance/Computed Tomography Imaging Software Application (referred to as
cvi42) is a software as a medical device (SaMD) that is intended for evaluating CT and MR images of the cardiovascular system.
Combining digital image processing, visualization, quantification, and reporting tools, cvi42 is designed to support the physician
in the evaluation and analysis of cardiovascular imaging studies.
Project 2
The TruPlan Computed Tomography (CT) Imaging Software application (referred to as “TruPlan”) is a software as a medical device
(SAMD) that helps qualified users with image-based pre-procedural planning and post-procedural follow-up of the Left Atrial
Appendage Closure (LAAC) procedure using Digital Imaging and Communications in Medicine (DICOM) images from CT
scanners.
All those project that keeps evolving and getting new functionalities that need to be continually tested by the QA team member and
the funding through this BSI internship will enable us to have a Quality Assurance (QA) team member to work with both the
platform development team as well as the R&D team. The idea of switching the intern through those project is to expose them
to different technologies and test strategies so they can learn as much as possible in the period they will be collaborating with
us.

Patient’s knowledge, attitudes and practices related to Human papillomavirus (HPV) vaccines: A systematic literature review.

The development of human papillomavirus (HPV) vaccine led to a reduction in the incidence of cervical cancer in women. But in low- and middle-income countries, the adoption of HPV vaccine is low. Hence why women are more likely to develop cervical cancer in those areas. There is a lack of understanding of the patient’s point of view regarding prevention by vaccination. This may have an impact on the implementation of HPV vaccination programs. By performing a systematic literature review with the patient’s perspective, we will be able to identify the limits regarding the optimal adoption of the HPV vaccine. Through this internship, the partner organization can participate in knowledge dissemination.

Predictive dosimetry in radiopharmaceutical therapy using advanced artificial intelligence, physiologically based pharmacokinetic modeling, and imaging data

Cancer is a leading cause of death worldwide. One of the main ways to remove cancer cell is targeting them using beta and alpha particle radiation. These kinds of radiations can be reached to the cancer sites by attaching them to special drugs that have specific binding sites on their cells. However, there is no accurate method to find how much radiation dose is required to remove these cells. As such measuring accurate dose received by cancer cells is not a straightforward method. In this era, we need to personalize therapy in terms of prediction of the dose. In this study, we try to predict dose using AI and imaging data, as well as models that describe the behavior’s of drugs in the body.

Generation and characterization of induced pluripotent stem cell (iPSC) lines from genetically engineered mouse model and differentiation potential

Tissue-resident fibroblasts and their attendant activities are central drivers of a diverse range of diseases. The overarching theme of this proposal is to develop novel fibroblast reporter lines that will enable us to identify
tractable therapeutic targets to modify fibroblast activity. To identify such targets, we are proposing to carry out high throughput screens (i.e., small molecules, gRNAs, siRNAs, etc.) in fibroblasts engineered to express a
reporter reflective of the desired phenotype. For the purposes of this proposal, we plan to generate induced pluripotent stem cell (iPSC) lines from fibroblasts isolated from a genetically engineered mouse model (GEMM).
Fibroblasts from these mice contain two reporter genes that will enable us to follow various aspects of the fibroblast phenotype. The project contains two aims: 1) generation and validation of iPSCs from mouse embryonic
fibroblasts of a GEMM; 2) generation of fibroblasts with the desired phenotypic properties from the aforementioned iPSC lines. Once validated, these iPSC lines will serve as “a well-validated, off-the-shelf” supply of fibroblasts for screening purposes. In short, this project will generate new iPSC cell lines to support a unique platform directed towards developing fibroblast-targeted therapeutics to treat disease.

Vers une amélioration continue de la gestion des systèmes de qualité en industrie pharmaceutique: l’informatisation des processus la bonne alternative de la documentation

Dans l’industrie pharmaceutique, la qualité des médicaments est de plus haute importance car elle touche des enjeux plus larges en termes de santé publique. Alors une démarche qualité est indispensable car elle est considérée comme un pilier de l’entreprise.
Les organismes pharmaceutiques ont tendance de mettre en place un système de management de qualité qui est l’ensemble des activités par lesquelles ils définissent, mettent en oeuvre et revoient leur politique et leurs objectifs qualité conformément à leur stratégie, et celui-ci est nécessaire à la maitrise et l’amélioration des divers processus.
Alors, dans le but d’avoir l’excellence opérationnelle dans la gestion quotidienne des systèmes qualité, on vise dans ce projet à optimiser les systèmes qualité en mettant en place des outils plus fiables qui rendent les processus plus performants. L’informatisation est l’une des principaux moyens qui permettrait des systèmes beaucoup plus flexibles et réactifs, d’autres approches d’optimisation de la gestion de la qualité vont être discutés dans ce projet tels que la cartographie des processus et l’établissement des KPIs.

Zero-to-Hero: Data Augmentation with LLMs and Human Feedback

Intent classification is a highly popular task within research communities and industries. Having a system that can classify intents of emails or messages from users can lead to a wide range of applications such as ticket routing and issue resolution. However, training such models require a large set of labeled data that might not be readily available. In this project work, we aim to present a system where only a few examples are required to be given for each intent. At each cycle we train an intent classifier and use a large language model (LLM) to generate extra examples using the examples in the dataset. Those generated examples are carefully selected as humans are involved to verify them. Concretely, the LLM generates a set of examples that are ranked based on scores given by the models involved in the system. To have more diversity and variations in the generated examples, we allow the LLM to generate new examples using previously generated examples. This system has the potential to train a powerful intent classification method with extremely low human effort that could significantly speed up the process of addressing customer requests.

Omy Laboratoires : Segmentation clientèle

Omy Laboratoires vise à offrir des soins topiques responsables, sains et personnalisés à ses client.e.s et
futurs client.e.s, à chaque étape de leur vie. Dans cette perspective de soins personnels et personnalisés,
Omy cherche à améliorer la compréhension de sa clientèle courante et à venir, en étudiant les différents jeux
de données collectés sur sa plateforme d’analyse d’images cutanées. Plus particulièrement, Omy cherche à
déterminer les caractéristiques déterminantes de la pérennité de sa clientèle et les segments les plus propices
pour sa croissance.
Dans ce contexte, OMY vise :
1. à expliquer ses performances de rétention de sa clientèle en déterminant les caractéristiques des client.e.s
qui persistent et les actions à entreprendre qui maximise cette rétention;
2. à déterminer les caractéristiques des différentes clientèles potentielles à développer en priorité.
Pour ce faire, OMY vise à étudier les données collectées lors des premiers contacts clients et leur première
évaluation dermique automatisée. Une attention particulière a été portée à ce que l’étudiant n’accède pas aux
images des utilisateurs mais bien aux caractéristiques extraites de ces images par des algorithmes
développés par un autre partenaire.