Innovative Projects Realized

Universal Graph Embeddings for Transaction data

Here at Mastercard, we have a transaction data where an account (card holder) transacts at a merchant, this data can be viewed as a Bipartite Graph of account and merchant nodes, where each edge represent a transaction between them. We have multiple models/tasks that provide predictive intelligence at an account level, e.g., predicting an account likely to experience a fraud, predicting an account likely to raise a chargeback, and predicting an account likely to default in near future. Goal of this work is to learn universal embeddings of account and merchants that can be leveraged to improve multiple downstream models.

Developing a Computational Model of Blood Clot Contraction to Predict Hemostasis and Wound Closure

Severe hemorrhage is a leading cause of mortality, accounting for upwards of 100,000 annual deaths in North America and 1.9 million worldwide. Under this condition, in a process called hemostasis, platelets in blood aggregate to form a plug to seal the blood vessel wound. During hemostasis platelets also contract to strengthen this plug, thus generating blood clots. Although substantial knowledge of the biology and biochemistry of clot contraction has been developed, the physics and mechanics of the process remain poorly understood. It is also unclear whether clot contraction aids in pulling together and sealing the wound. The goal of this project is to develop a multi-scale computational model able to predict clot contraction and blood vessel wound closure. The model will be based on the mechanics of clot contraction and its underlying microscopic biophysical principles, validated with experimental data from modified in vitro platforms. A deeper understanding of the mechanisms regulating clot contraction can lead to the development of novel approaches in the treatment of bleeding and clotting disorders.

Alternative approaches to curing flip chip underfill encapsulants

The volume of data being transmitted (streaming, blockchain, machine learning, etc.) continues to increase exponentially. To carry this data, the traditional metallic transmission systems (typically copper) are being replaced with light-based (photonic) medium to reduce both the space and power consumption. This implies integrating photonic/optical components within the microelectronics infrastructure for electrical to optical conversion and optical interconnection. A key area of focus to bring such technology to the mainstream is the connection of multiple optical fibers to the microelectronic package using a photonic integrated circuit (PIC) as the interface. To accomplish such fiber connections at high densities yet in a highly automated, cost-effective fashion, the PIC chip must be maintained extremely flat in the region of connection, yet the material sets and processes inherent to the microelectronic package are conducive to stresses that bend (warp) both the package and the chips therein, including the PIC.

Étude et optimisation de biocapteurs plasmoniques par des effets de thermofonctionnalisation

Le projet porte sur l’optimisation de biocapteurs grâce à une nouvelle méthode utilisant un laser impulsionnel (excitation très courte). Les deux étudiants du projet travailleront sur deux aspects distincts du projet: d’un côté, l’étude de matériaux aux propriétés thermiques et optiques intéressantes pour lesquels nous avons peu de données; d’un autre côté, l’étude de l’effet d’un laser pour enlever des molécules d’une surface et optimiser leur position sur le capteur. Ce travail conjoint devrait permettre d’obtenir des biocapteurs avec des nanostructures qui seront plus performants que ceux actuels.

Improving the Crystal Morphology of Rufinamide by Experimental Structure Determination and Advanced Computational Approaches

Crystal shape and size are among the most important features of crystalline product that are prevalent in the pharmaceutical industry. Extremely fine and needle-like crystals pose various problems in the downstream processes including filtration, tableting, handling, and storage. Rufinamide is an anticonvulsant medication, and its most stable and therapeutically advantageous crystalline form shows notoriously undesirable morphology (filiform or thread-like) and is selected as a model compound for this study to change its morphology to a more equidimensional shape.

We approach this challenge by obtaining the atomic resolution of crystalline Rufinamide by single-crystal X-ray diffraction and identifying functional groups exposed on each crystalline surface. Using the obtained information from the crystal structure, we can use tailor-made additives to hinder the growth of the long axis and induce more equidimensional crystal formation. Computational approaches will also be performed on the obtained crystal structure, which provides a thermodynamic understanding of why needles are formed and how we can hinder their natural growth process by selected additives. The procedure developed in this research will provide valuable experimental data and theoretical understanding of controlling the morphology of pharmaceuticals.

Predicting falls based on a 2-minute walk test

Falls are the leading cause of injuries in older adults. Identifying older adults with risk for falls prior to discharge home from the Emergency Department (ED) could help direct fall prevention interventions, yet ED-based tools to assist risk stratification are under-developed. The aim of this study was to compare the performance of our proposed machine learning algorithms with existing screening tools to predict future falls in the 90-days post ED discharge for 150 older adults aged 65 years and older.

Développement et évaluation des performances d’outils pour la sélection des emplacements d’échantillons instantanés dans des réseaux de distribution d’eau potable réels

Avec l’aide de la technologie développée par SWDS, les services publics d’eau potable pourront lever plus tôt et à moindre coût les avis reliés à l’eau potable, tels que les avis d’ébullition de l’eau ou de ne pas boire d’eau. Lors d’un incident de contamination de l’eau potable, les outils de SWDS peuvent trouver les zones touchées plus rapidement et plus précisément que les outils existants. Avec les outils de SWDS, l’injection de désinfectants, comme le chlore, sera ajustée en fonction de la consommation d’eau réelle plutôt que sur la base d’un calendrier typique d’injection. SWDS fournit tous ces services aux services d’eau en réalisant un jumeau numérique fiable du réseau de distribution d’eau potable pour améliorer la sécurité et la qualité de l’eau potable.

Amination of redox-active esters of fluorinated alicyclic carboxylic acids with diazirines

The objective of this research program is the development of new green synthetic methods with emphasis on metal-free procedures for the simple preparation of biologically and commercially important molecules. In particular, fluorinated sp3-enriched Nitrogen-containing compounds will be prepared by radical reactions of redox-active esters (i.e., N-hydroxyphthalimide, N-hydroxy-2-thiopyridone, etc.) derived from fluorinated alicyclic carboxylic acids with the diazirine derivatives.
According to this, we are going to establish reactivity patterns for the redox-active esters derived from fluorinated alicyclic carboxylic acids in the reaction with diazirines. Then find optimal reaction conditions for this transformation, providing either imine or diazirine products selectively and demonstarate possible applications of the products obtained for the synthesis of potentially useful fluorinated Nitrogen-containing compounds, e.g., amines, hydrazines, or nitrogen-containing heterocycles.

Dual Eye-tracking for Measuring Surgical Team Cognition

Team collaboration is important for performing image-guided surgery, including laparoscopic surgery. Therefore, training
healthcare providers in a team setting is critical for the success of laparoscopic procedures.
To improve the quality of any team training program, we need to develop a reliable assessment instrument to measure team
performance. The currently available team assessment tools are mostly subjective and indirect. What we need is to find a way to
directly evaluate team cognition, the psychomotor foundation behind any team performance.
We have used eye-tracking technology to assess the cognition of an individual surgeon. In this study, we plan to develop and test
a dual eye-tracking technology and further investigate whether eye-tracking can be used to measure team cognition between two
surgeons in a laparoscopic surgery. The task goal will be fulfilled by building the collaboration with scientists and surgeons in the
Institute of Image Guided Surgery (IHU-Strasbourg) in France. One intern (Yao Zhang, PhD student) will be supported for 12
weeks by GRA fund.

Étude de marché pour la création d’une compagnie de photographie aérienne

Le projet consiste à relever les défis liés à l’étude de marché pour la création d’une compagnie de prise de vue
aérienne via les différentes technologies innovantes et la validation du plan d’affaire pour la fondation de cette
entreprise qui feront l’objet de ce programme.
Le but du projet est l’optimisation et la validation du potentiel commercial de l’idée afin d’affiner le concept et de
combiner les différentes technologies en la matière qui seront les atouts nécessaires pour la création de
l’entreprise.

Couling Inventory Validation

Cultural Planning encourages recognition of a community’s network of interrelated assets, and maintains that it is beneficial to conserve built cultural heritage resources. Many municipalities, including the City of Guelph, have undertaken cultural mapping initiatives, which are considered an initial step and a defining feature of this vein of planning. Guelph’s mapping initiative identifies the location of a variety of assets including buildings and structures; however, several non-designated or non-listed historic properties are currently not included. These resources are acknowledged in an inventory of over 2000 properties created by Gordon Couling – a notable figure in Guelph’s heritage conservation movement. My project will assess the validity of Couling’s inventory, maintaining a specific focus on architectural and contextual value as defined by the criteria for determining cultural heritage value or interest provided by Ontario Regulation 9/06 within the Ontario Heritage Act (OHA). This research will assist the City of Guelph in deciding if they should pursue formal protection for individual resources, or if concentrations of significant properties represent potential heritage conservations districts.

Beyond “proof of concept” – machine learning and related tools applied to transactional data extracts from clinical and business information systems – in service of Quality Assurance/Quality Improvement.

The program of eight interlocking applied research projects employs various machine learning (ML) methodologies with large volumes of Island Health cross-continuum patient-service encounter data inside the Island Health firewall. The objective is to generate ready-for-use products that will promote better outcomes via more effectively optimized interoperation of IH programs. As well, the work targets objectives of health system sustainability by identifying, counting and costing out processes that are difficult to reconcile with established clinical care guidelines. Complementing more usual diagnostically-based approaches, the project seeks to classify persons in terms of cross-continuum patterns of service utilization (PSUs), relate PSUs to outcomes, and employ PSUs as foundational units for tools that can be employed to build out evidentiary foundations for Quality Assurance/Quality Improvement targeting and approaches. Information products are expressed in an intuitive manner, to provide direct support for translation of generic service system strategies into tactically focused Quality Assurance/Quality Improvement initiatives.