ResearchMatch v2 initial submission

ResearchMatch - second iteration initial projects submissions

Please find below the list of 51 ResearchMatch projects received from the McGill life science and clinical communities. The McGill data science community is now being asked to contact the individual researchers if interested to collaborate in order to submit a full proposal by April 24, 2020.


Projects are split into 3 groups:

Clinical Research (18)  Life Science Research (22)  Population Health Research (11)


Clinical Research

Project title: A practical problem in need of a solution: preparing radiological data for clinical research on AI

AI-based analysis of radiology images is rapidly but there is a gap between how this technology is evaluated in computer science literature and the kind of evaluations that are clinically relevant. A common bottleneck in evaluative research on AI for analyzing radiology images is in the de-identification of images (e.g. chest X-rays, CT scans) and linking of images to clinical data using some form of unique identifier. I am looking for a collaborator to develop a user-friendly program that can accomplish this.

Keywords: radiology, evaluative research, tuberculosis, deep learning, AI, artificial intelligence

PI: FAIZ AHMAD KHAN (Associate Professor)
Department: Communication Sciences and Disorder



Project title: Development of a Personalized Pain Treatment Index

This proposal attempts to fill a gap in the way we manage pain: 1) The current tools of measuring pain do not take into account the individual pain thresholds influenced by the neurophysiology of an individual. They also don’t distinguish the underlying causes of pain or the different pain sensations; 2) This leads to the idea that any painkiller medication (including opioids) will “kill” all pain. Part of the problem is that physicians have no reliable way of predicting how any particular patient will respond to painkillers. We will develop a Personalized Pain Treatment Index (PPTI) based on clinical predictors of poor pain control that will inform treatment decision-making. The central hypothesis of this proposal is that the PPTI will help in guiding the pain treatment and allow for individualized treatment of children and adolescents with chronic MSK pain.

Keywords: pediatrics, sensory testing, personalized treatment, psychophysical profiling, pain.

PI: Catherine Ferland (Assistant Professor)

Department: Anesthesia

email: catherine.ferland [at]


Project title: Predicting response to anti-TNF therapy in Rheumatoid Arthritis

Rheumatoid arthritis (RA) is a complex, chronic disease of the immune system characterized by disfiguring and disabling arthritis. Tumour necrosis factor (TNF) inhibitors have revolutionized the treatment of RA in the last decade. However, the fact is that 30% of patients will fail to respond to TNF inhibitors and we currently have no reliable way of predicting response. In 2013, Sage and DREAM (Dialogue for Reverse Engineering Assessments and Methods) launched the Rheumatoid Arthritis Responder Challenge (RARC) to identify genetic predictors of response to anti-TNF treatment in RA ( Despite finding a significant genetic heritability estimate of a treatment non-response trait, no significant genetic contribution to prediction accuracy was observed in any of the models submitted by 73 research groups. We wish to re-analyze the data from DREAM RARC using alternative strategies for identifying SNPs of interest, including novel methods using functional enrichment to improve polygenic risk prediction accuracy.

Keywords: Polygenic risk scores, Rheumatoid arthritis

PI: Marie Hudson (Associate professor)

Department: Medicine

emailmarie.hudson [at]


Project title: Leveraging health services data with genomic data in early stage lung cancer  

In Quebec, lung cancer (LC) is the leading cause of cancer death in both men and women. The prevalence of early stage non-small cell lung cancer (NSCLC) is expected to increase given current trends in population aging and the advent of low dose CT screening. The MUHC is the regional Quebec ministry-mandated lead for LC and coordinates care for over 1100 new LC patients per year. Traditionally, clinicians integrate clinical data to assess fitness for all available therapeutics and the likelihood of a good oncological outcome, balanced with optimal quality of life. Such clinical data may include pulmonary function, blood tests, comorbidities, imaging features and molecular pathology results, among others. However, in today’s modern clinic, the wealth of data we can potentially identify is too large for any one clinician to aggregate and use for clinical decision making. This includes 1) medical records and claims data from the MUHC Data Warehouse; 2) imaging data from CT scans using advanced computerized image analysis approaches such as radiomics; 3) pathology and molecular diagnostics;4) genomics data from lung cancer specimens identified through next generation sequencing, and 5) pulmonary function data performed pre-operatively on patients evaluating the physiologic properties of the lung. Our aim is to integrate these data sets into one location so they can be used for risk prediction of post-operative outcomes and survival of lung cancer patients treated by lobectomy at the MUHC. 

Keywords: data warehouse, genomics, Big data analysis, lung cancer, radiomics

PI: Nicole Ezer (Assistant Professor)

Department: Medicine

email: nicole.ezer [at]


Project title: Sex specific biomarker expression in patients with type 2 diabetes mellitus

Diabetes mellitus (DM) is a major public health crisis in Canada: by 2030, over 1 in 5 individuals are projected to develop DM resulting in total costs to the healthcare system of $336 billion. Overall, DM increases the risk of cardiovascular (CV) death by approximately 38%, all-cause mortality by 40% and hospitalization by 33%. Across biologic sex (defined as male or female), the burden of CV outcomes (including HF and myocardial infarction) among patients with DM appears to be significantly higher compared to patients without DM. Furthermore, as we have previously described, DM is one of the most prevalent comorbidity among individuals with established CV disease regardless of biologic sex. Across multiple CV specialities, there is limited but increasing recognition of the differences in the sex specific distribution of CV outcomes. However, there are major knowledge gaps in our understanding of the distribution of specific causes of death, predictors of causes of death, and mechanism contributing to these causes of death among individuals with DM across biologic sex. We propose the ‘identification and prediction of Sex-specific Causes of Death in patients with Diabetes Mellitus post Myocardial Infarction’ (iSCD-DM-MI) Study to address these critical knowledge gaps. 

Keywords: diabetes , phenotypes , biomarkers , heart failure , myocardial infarction 

PI: Abhinav Sharma (Assistant professor)

Department: Cardiology

email: abhinav.sharma [at]


Project title: Predicting Emergency Department Overcrowding Using Machine Learning Techniques

Emergency Department overcrowding is not only dangerous for our patients but a serious threat to the physical and psychological wellbeing of our staff (doctors, nurses, orderlies, support staff). While we can adapt to gradual trends in patient volume over longer periods, the number of daily ED patient visits can vary greatly from one day to another. Using machine learning, deep learning, and autoML techniques on live data, I am developing predictive models of patient visits and hourly overcrowding rates, with the goal of eventually adjusting staffing proactively before the situation becomes dangerous. These models could easily be deployed live for all EDs across Quebec in order to reduce overcrowding and deliver better and more timely care to our patients. A live prototype of one of my models:

Keywords: emergency, emergency medicine, overcrowding, time series forecasting, emergency department overcrowding, machine learning, artificial intelligence, AI, deep learning, statistical analyses, Big data analysis

PI: Devin Hopkins (Emergency Physician, Clinical Lecturer)

Department: Emergency Medicine

email: devin.hopkins [at]


Project title: MARVIN: Developing an Intelligent Conversational Agent to Promote HIV Patients’ self-Management

Patients with chronic conditions such as HIV infection play an important role in managing their conditions. Intelligent conversational agents (ICA) are proven to be cost-effective tools to improve adherence related barriers in chronic health conditions and to promote patient empowerment, collaborative goal settings, and problem-solving skills. Therefore, the project objective is to design, develop, and test an ICA to enhance patient self-management in care with a specific focus on the improvement of barriers to ART adherence. The project has already started in January 2020 and first prototype of Marvin is currently built by a group of software students from Polytechnique Montreal with Rasa platform. Currently, we are looking for highly qualified personnel in order to train the AI layer of the Marvin solution using our Q&A corpus and to continue to improve both Marvin’s functionality and quality of conversation. Additionally, this project will provide further opportunities for cooperation with McGill data science community to adapt ICA for patient self-management with other chronic conditions.

Keywords: HIV infection, AI, Patient-oriented research, self-management, ehealth, Intelligent Conversational Agent

PI: Bertrand Lebouche (Associate professor)

Department: Family Medicine

email: bertrand.lebouche [at]


Project title: HLA Epitope Compatibility: A Strategy to Prevent Immune Injury in Transplantation

When a donor kidney is transplanted, the recipient’s immune system may recognize it as incompatible and reject it. We seek to promote lifetime kidney transplant survival by applying novel computational tools, which can verify optimal donor-recipient compatibility at the level of human leukocyte antigens. This strategy can allow patients enjoy their grafts for their lifetime as well as reduce immune suppression medication needs with their associated complications.

Keywords: statistical analyses, immunology, clinical prediction, complex systems, data mining, disease cohorts, disease surveillance, infection and immunity, precision medicine, deep learning, artificial intelligence, machine learning, Big data analysis

PI: Ruth Sapir-Pichhadze (Associate Professor)

Department: Medicine

email: ruth.sapir-pichhadze [at]


Project title: Neuroplastic Effects of Transcranial Stimulation in Pre-Symptomatic Cognitive Impairment

Transcranial direct current stimulation (tDCS) is a safe, non-invasive and cost-effective method to improve cognition by applying weak electric currents to the scalp. Applying tDCS during the earliest signs of memory loss, such as subjective memory complaints, may be help prevent further declines. Yet, findings are inconsistent and the neural processes through which tDCS influences cognition and behaviour are unknown. Crucially, few tDCS studies have considered individual factors, which may influence tDCS responsiveness through biological, cognitive, or psychological processes that draw on similar neural networks as memory and related cognitive processes. This study evaluates the viability of tDCS for treating pre-symptomatic cognitive impairment and explores how tDCS affects the intrinsic functional connectivity of the aging brain, taking into account individual factors that may alter the neural mechanisms underlying tDCS effects.

Keywords: brain research, Image analysis and machine learning, clinical prediction, Patient-oriented research, imaging

PI: Sheida Rabipour (Postdoctoral fellow)

Department: Psychiatry

email: sheida.rabipour [at]


Project title: The development of a predictive model of attrition for pediatric weight management care

In response to the burden of obesity among children and youth, pediatric weight management programs have been implemented. However, retention in these programs is typically low, with up to 3 out of 4 participants dropping out before program completion. Using a clinical database of children and youth enrolled in a Montreal-based weight management program, this project aims to develop a predictive computational model of attrition using individual, familial and neighbourhood-level data. The ability to identify, at program entry, which participants are most likely to drop out of the program will allow for targeted retention strategies to be implemented at the outset.

Keywords: children/adolescents, neighbourhood environment, family environment, data mining, pediatrics, personalized treatment, obesity, clinical prediction

PI: Andraea Van Hulst (Assistant Professor)

Department: Ingram School of Nursing

email: andraea.vanhulst [at]


Project title: The use of deep learning to identify disrupted brain oscillatory patterns of aging subjects

The rapid growth of the aging population is an alarming cause for concern. The prevalence of disability rises with age, increasing the costs on the healthcare system. The aging process involves changes in the function of the brain, which can be quantified via neuroimaging techniques such as magnetoencephalography (MEG). These changes are associated with behavioural deficits in both the cognitive and the motor domain. One way to look at functional changes in the brain is to measure phase-amplitude coupling (PAC, amplitude coupling of different frequency bands), which is thought to reflect processes underlying information gating and communication across brain networks. PAC has been shown to be disrupted in aging. It has also been shown that aberrant PAC can be rescued by non-invasive brain stimulation (NIBS) techniques. More specifically, transcranial alternating current stimulation (tACS) has been shown to have great potential to entrain brain networks in a frequency-specific manner with lasting effects on cognitive functions. This suggests that tACS targeting specific frequency band coupling, could also be beneficial to motor performance. Here we proposed to use deep learning (DL) to identify disrupted brain oscillatory patterns in the motor network of aging subjects. To do so, we request the help of someone with expertise in DL in order to validate and optimize the use an algorithm we have developed in the context of analysing another small neuroimaging dataset. Ultimately, the results of this study will guide the formulation of hypotheses on how to best target and normalise disrupted brain activity using NIBS. Our long-term goal is to implement real-time monitoring of brain activity and guide the design of individualised interventions for normalizing brain patterns/interactions in order to promote motor/cognitive performance. 

Keywords: motor performance, aging, phase-amplitude coupling, deep learning, neuroimaging

PI: Marie-Hélène Boudrias (Assistant Professor)

Department: School of Physical and Occupational Therapy

email: mh.boudrias [at]


Project title: Facilitation of assessment for speech and language disorders via automatic speech recognition

Speech and language disorders are very common in children and adults but heterogeneous in their underlying cause. Many developmental conditions and neurological impairments cause difficulties with speaking clearly, understanding language accurately or communicating effectively. Individuals with these speech and language disorders have poorer quality of life due to reduced opportunities for social inclusion, academic success and vocational attainment. Effective treatment of these disorders by speech-language pathologists requires accurate diagnosis of the underlying cause, for example, auditory processing impairments, memory deficits or speech motor control issues. A simple but powerful test helps the speech-language pathologist to identify the cause and design an effective intervention: the patient is asked to listen to strings of nonsense syllables and repeat them. Although these nonsense syllable tests are easy to administer, they are time consuming and difficult to score. We propose to develop a digital application to administer the test and obtain the patient’s responses. Furthermore, we propose to incorporate automatic speech recognition technology to score the patient’s speech responses. This type of digital application will increase clinical use of this assessment technique and improve the efficacy of intervention for large numbers of children and adults with speech and language disorders. 

Keywords:  digital applications for health, phonetic transcription,  assessment and intervention of language disorders, non-sense syllable repetition/non-word repetition, automatic speech recognition

PI: Aparna Nadig (Associate Professor)

Department: School of Communication Sciences and Disorders

email: aparna.nadig [at]


Project title: Developing a Simple Bedside Clinical Estimator of Extubation Readiness in Extremely Preterm Infants

Extremely preterm infants (born before 28 weeks gestation) are a highly vulnerable population that commonly requires intubation and mechanical ventilation after birth. Due to various complications associated with prolonged mechanical ventilation, clinicians try to extubate these infants as early as possible. However, in current practice, the decision to extubate is highly subjective and varies significantly from one clinician to the other. As a result, many infants are exposed to mechanical ventilation for unnecessarily prolonged periods of time, while others are prematurely disconnected from the ventilator and reintubated. For those reasons, it would be ideal to develop a simple clinical bedside tool that could accurately estimate an infant’s probability of having a successful or failed extubation at any given day while mechanically ventilated. Using data obtained from the largest prospective multicenter study about extubation, we propose applying machine-learning methodologies to develop a simple clinical estimator of extubation readiness in extremely preterm infants.

Keywords: artificial intelligence, risk prediction, extubation, extremely preterm infants, mechanical ventilation, precision medicine, machine learning

PI: Wissam Shalish (Neonatologist)

Department: Pediatrics

email: wissam.shalish [at]


Project title: Role of telomere length in Parkinson's disease; genomic and mendelian randomization study

Parkinson’s disease (PD) is a common neurodegenerative disorder with complex pathogenesis. Out of all the risk factors, age is the strongest risk factor for PD development. Genetics also play significant role in PD with variable rate of heritability. Mutations in GBA, encoding for the lysosomal enzyme glucocerebrosidase, have been identified as the most common genetic risk factor of PD. Other genes, related to lysosomal metabolism (TMEM175, SMPD1) have also been associated with PD. Telomeres play a crucial role in the sustaining of genome function. Recently, 17 loci have been associated with telomere length (TL) in a genome wide association study (GWAS). TL has been associated with neurodegeneration and age-related diseases such as Alzheimer disease. Moreover, TL has been linked with early aging in Fabry disease, a lysosomal storage disorder. In the current study, we propose to analyze the role of the loci associated with TL in PD using available large-scale genomic data of 20,000 individuals (PD patients and controls). We will study phenotypic effects of TL on PD, including possible modification of age at onset. To find whether TL may be causal in PD, we will perform a Mendelian randomization study using the available GWAS data on TL and PD. We will also use large scale data from UK Biobank (> 450000 participants) to study the role of TL in people with available clinical phenotype. Thus, by utilizing existing genetic and clinical data on large cohorts of PD patients, we will identify role of TL in PD.

Keywords: Big data analysis, genetics, genomics, precision medicine, uk biobank, GWAS, Genomics, Human genetics

PI: Ziv Gan-Or (Assistant Professor)

Department: Human Genetics

email: ziv.gan-or [at]


Project title: Fatty Acids and Aortic Stenosis: Genetics, Metabolic Biomarkers, and AI for improved Prediction

Aortic stenosis (AS), the most common form of heart valve disease, has no known treatments to slow its development. AS impedes blood flow and forces the heart to work harder to pump blood and leads to heart failure and death. With the aging population, AS prevalence is projected to increase by >2-fold by 2040, and by 3-fold by 2060. Despite this large burden, the etiology of AS remains poorly understood. An increased understanding of the causes of AS is needed to identify avenues for prevention and treatment. Genomic data can be used for drug target identification because individuals with naturally occurring risk genotypes act as models of drug therapy, permitting the discovery of therapeutic pathways. Single nucleotide polymorphisms (SNPs) identified by genome-wide association studies (GWAS) have demonstrated enrichment for known drug targets. Recently, we identified the locus, FADS1/2/3 as a novel pathway for AS. However, we need to identify the precise biological target that mediates the association of FADS enzymes with AS. The FADS1/2/3 gene locus is a cluster of fatty acid desaturases that strongly effects fatty acid metabolism. We hypothesize that the variant reduces inflammation, a causal pathway for AS. However, while the proximal effects of the FADS1/2/3 locus are well known, the downstream effects are less well understood. Using fatty acid assays combined with lipidomics, we aim to identify markers of disease risk and specific therapeutic targets in plasma samples. This will be achieved by developing novel models from artificial intelligence to improve the utility of genetic profiling.

Keywords: aortic stenosis, artificial intelligence, machine learning, clinical prediction, cardiovascular disease, GWAS, CLSA

PI: James Engert (Associate Professor)

Department: Experimental Medicine

email: jamie.engert [at]


Project title: Development of a liquid biopsy for monitoring immune checkpoint inhibitors associated toxicities

Checkpoint inhibitors (CPIs) reinvigorate cancer surveillance by blocking negative immune checkpoints that have been hardwired into our immune system to maintain homeostasis and prevent autoimmunity. On one hand, the advent of CPIs has led to remarkable cancer remissions in tumors previously refractory to all standards of care treatments. On the other hand, the use of CPIs can also lead to a remarkable spectrum of autoimmune reactions, named immune-related adverse events (irAEs), which can affect any of a hosts’ organs. irAEs can be associated with significant morbidity, mortality, and prolonged hospitalizations. To date, no test or biomarker can reliably predict which patients receiving CPIs will suffer from irAEs. Furthermore, the soluble mediators of irAEs are largely underexplored and heavily extrapolated from primary models of autoimmunity. Our group’s preliminary data, obtained from multiplexed protein measurements on commercial assays from MSD and Luminex, suggest that some protein levels have a correlation with irAE. However, these closed technology platforms do not permit the proteomic pathways involved in CPI to be investigated in more detail, since new targets would cross-react with the existing panel. Importantly, the nELISA platform, developed by nplex, is open and readily expandable to new targets, owing to the cross-reactivity free nature of the platform. Thus, as a first step towards better understanding CPIs, irAEs, and the development of clinical diagnostics for irAEs, we propose the development of a CPI-specific nELISA panel in collaboration with nplex biosciences Inc. We will build a 30-plex nELISA panel specifically focused on CPI-related proteins, and use this panel, in combination with nplex’s existing 100-plex nELISA panel (Specific to chemokines and cytokines), to discover preliminary biomarkers of irAEs. After quantitatively profiling the levels of 130 proteins in 200 patient samples (already biobanked), we expect to uncover the first preliminary clinical biomarker for CPI irAEs, and in parallel, potentially uncover new CPI biology. If successful, (i) this project would represent a massive step forward in our understanding of irAEs, (ii) firmly demonstrate the potential of nplex’s nELISA platform for clinical biomarker discovery, and (iii) spur the further commercial development of the first clinical diagnostic for irAEs based on an nELISA protein signature. 

Keywords: Cancer immunotherapy, Immune-related adverse events, Immune checkpoint inhibitors

PI: Khashayar Esfahani (Assistant Professor)

Department: Oncology

email: khashayar.esfahani [at]


Project title: Computational Modeling of the Human Heart Ventricles: Hypertrophy and Dilation

Heart diseases are the leading cause of mortality and morbidity in developed countries. It is estimated that almost 5 million North Americans develop heart conditions leading to heart failure. Heart failure (HF) is a syndrome of ventricular dysfunction whereby the damaged or weakened heart cannot supply sufficient blood flow to the body.Computational modelling of the cardiovascular system using patient-specific information will provide an effective, and more importantly non-invasive tool to cardiovascular specialist using which they can offer better diagnosis and more successful treatment to patients with cardiovascular diseases. In order to understand disease progression,quantitative prediction of wall stress and strain states in cardiac tissues are of paramount importance. It is thus vital to determine clinically the critical point at which irreversible changes in the myocardial fibers necessitate intervention. In this study an attempt is made to investigate cardiac ventricular dilation using combined patient-specific finite element analysis and statistical techniques. The first goal is to propose a methodology, using machine learning methods, to automatically categorize patients in different stages of heart failure using cardiac images. The second goal is to propose finite element and statistical methodologies to predict progression of heart failure for each patient

PI: Dominique Shum-Tim (Professor)

Department: Surgery

email: dshumtim [at]



Life Science Research

Project title: Regulation of axon-myelin relationships: Insights from gene-edited mice

Myelin sheaths are composed of spirally wrapped glial cell plasma membrane tightly compacted by a family of myelin-associated proteins that protect neuronal axonal fibers. They increase the transverse resistance of each fibre and lead to voltage-gated Na channel clustering at nodes of Ranvier supporting node-to-node saltatory conduction, which vastly accelerates conduction rates and limits repolarization energy requirements to the nodal domain. Adult myelin is adaptive with rapidly growing evidence suggesting that such plasticity plays a key role in both normal and abnormal nervous system function. Insight into the mechanism through which such myelin changes are mediated remains limited.

Keywords: Complex systems modeling, Image analysis and machine learning, Big data analysis, G-ratio, Myelin-sheath thickness

PI: Alan Peterson (Associate Professor)

Department: Oncology

email: alan.peterson [at]


Project title: Large-scale gene-by-environment interaction (GxE) modelling using deep learning

Using deep learning, we seek to construct complex Gene by Environment interaction (GxE) models which could identify patients at risk of depression/ADHD and determine which modifiable aspect of the environment should be targeted for maximizing improvement (personalized treatment). We will use DREAM BIG, a consortium of 5 harmonized prenatal cohorts.

Keywords: genetics, imaging, statistical analyses, deep learning, artificial intelligence

PI: Ashley Wazana (Child psychiatrist researcher)

Department: Psychiatry

email: ashley.wazana [at]


Project title:  LabRemote

Experimental results are often not reproducible because metadata collection and annotation is difficult, time-consuming, and can be lost or altered. In addition, subtle differences in experimental protocols used by researchers are ubiquitous but difficult to quantify: even the same research group can struggle to reproduce results over time. This project aims to build an open source platform that improves research reproducibility using mobile technology. My laboratory has developed a prototype app that functions as a laboratory remote control while also time-stamping every aspect of a researcher's protocol, allowing us to capture complete metadata during experiments without any additional time commitment on the part of the user. A collaboration with the computer science department is needed to develop data mining and visualization tools to link experimental outcomes to variations in protocols, an audit trail to ensure that lab logs are unalterable, and to refine the apps voice capabilities to allow for hands-free operation and integration with existing mobile assistant technologies.

Keywords: Intelligent Conversational Agent, data mining, study design, machine learning

PI: Gil Bub (Associate Professor)

Department: Physiology

email: gil.bub [at]


Project title: Predicting the impact of CHD3 mutations based on missense tolerance and 3D protein structure.

CHD3 is an epigenetic regulator important for neurons, mutations in which cause a variable neurodevelopmental disorder with language delay and autism. The associated intellectual impairment can range from borderline to severe. With the increased use of genome and exome sequencing, new variants in this gene are found frequently in children with delays, and two subsequent questions clinicians and families have currently can't be answered; 1) Is the variants identified disease-causing?, and 2) can we predict the severity of the disorder associated with this variant? We hypothesize that by analyzing variant characteristics (whether the position is tolerant to variation in the population and throughout evolution, the type of amino acid change, and the 3D structure of the protein), we will be able to answer these questions with a high degree of confidence. We currently have detailed phenotype information of 40 mutations which we can use for training, the 3D structure of homologous proteins, different scores available for amino acid conservation and tolerance to variation. We are currently generating additional data (deeper neuropsychiatric evaluations, methylome analysis, recruitment of new patients) which should allow us to test our hypothesis. 

Keywords: quantitative medicine, Protein structure,  epigenetic, machine learning, pediatrics, precision medicine, genomics, genetics, brain research

PI: Philippe Campeau (Adjunct Professor)

Department: Anatomy and Cell Biology

email: pcampeau [at]


Project title: Profiling molecular steps in adipose tissue that lead to metabolic dysfunction in weight gain

Adipose tissue plays a central role in maintenance of healthy metabolism. In obesity, its capacity to expand via extracellular matrix remodeling, cell growth and lipid storage fails. This leads to ectopic lipid accumulation, whole-body low-grade inflammation, insulin resistance and ultimately to type 2 diabetes. This project examines the detailed profile of the extracellular matrix and inflammatory changes in adipose tissue that disrupt normal expansion during weight gain. The project uses whole adipose tissue transcriptome data set from a rare monozygotic, weight discordant twins as well as adipose tissue RNAseq data from longitudinal mouse weight gain study. Focus will be on adipose tissue matrisome alterations and changes in inflammatory profile during adipose tissue expansion. Ultimate goal is to identify molecular targets that would allow maintenance of healthy metabolism in obesity. 

Keywords: adipose tissue, transcriptome, type 2 diabetes, RNAseq, inflammation, extracellular matrix, obesity

PI: Mari Kaartinen (Associate Professor)

Department: Experimental Medicine / Medicine/Dentistry

email: mari.kaartinen [at]


Project title: Analyzing the contribution of genetic variants to familial 

The genetic factors predisposing to idiopathic scoliosis (where 30% have a family history of the disease), a condition affecting up to 3% of the population, are not well known. We've performed exome sequencing (sequencing all coding regions of the genome) in 40 families with multiple affected individuals. A list of candidate variants has been generated, but the candidate genes and variants need to be replicated in other cohorts. Data from other cohorts with scoliosis (Cartagene study in Quebec, and Gabriella Miller Kids First Program) can be obtained, and an analysis would be performed to determine if our candidate variant and gene list is applicable more broadly to other cohorts. 

Keywords: Exome sequencing, Rare variants, SKAT analysis, Population genetics, Genomics, Human genetics, Gene discovery

PI: Philippe Campeau (Adjunct Professor)

Department: Anatomy and Cell Biology

email: pcampeau [at]


Project title: Skeletal muscle adaptations to hypoxia and work in Zebrafish

Adult Zebrafish display excellent sustained swimming ability even under hypoxic conditions.  In contrast, in mammals,  including humans, hypoxia impairs muscle function and hypoxic lung and cardiac diseases are associated with loss of muscle mass and functional capacity.  This study was set up to investigate the transcriptional adaptations that occur in muscle of Zebrafish exercising under hypoxia.  It is hoped that the results will highlight new molecular targets that may prove useful to help preserve muscle mass and improve performance in human cardiopulmonary disease and systemic hypoxia. 

Keywords: Exercise, Zebrafish, Skeletal muscle, Hypoxia

PI: Thomas Jagoe (Associate Professor)

Department: Medicine

email: thomas.jagoe [at]


Project title: B cell immune profiling identify pathogenic B cell population for idiopathic nephrotic syndrome

We are investigating the pathogenesis of idiopathic nephrotic syndrome in children. The disease is featured by massive protein leakage in the urine and is believed to be caused by immune derangement. One of the effective treatment strategies is to deplete B cells but when the B cell counts recover, the disease tends to relapse. This suggests that subset(s) of B cells are pathogenic. We are conducting a pilot experiment using B cells isolated from four children when they are recovering from B cell depletion, one while still in remission and another after the disease relapses (i.e. paired samples from 4 patients=8 samples). The B cells are subjected for single cell RNA-seq both for whole transcriptome and specifically for B cell receptor repertoire. We are seeking for collaboration to analyze the data to assess the clonicity of B cells and the transcriptional changes that are likely to cause disease relapse. Depending on the results, we plan to expand our study.

Keywords: single cell RNA-sequencing, Nephrotic syndrome, B cell repertoire

PI: Tomoko Takano (Professor)

Department: Medicine

email: tomoko.takano [at]


Project title: Investigating intestinal stem cell regeneration at the host-parasite interface

Diseases affecting the digestive tract are commonly linked to intestinal barrier dysfunction and uncontrolled inflammation. As a result, the gut epithelium has evolved various defensive measures to counter invasive pathogens and quickly repair itself after injury. The remarkable regenerative capacity of the gut relies on the ability of both resident stem cells and specialised cells of the gut epithelium to radically alter their cellular identity, a phenomenon known as cellular plasticity. Understanding the gene networks that drive this process has the potential to lead to novel strategies for the prevention and treatment of inflammatory bowel diseases. Another avenue for future therapeutic intervention, lies in the emerging evidence that modulation of the microbiome may favorably orchestrate intestinal stem cell dynamics. Amongst the trillions of microorganisms inhabiting the gut, parasitic worms or helminths are surprisingly well tolerated due to their potent ability to dampen intestinal inflammation and promote tissue repair. Although most studies have focused on their immunomodulatory functions, in this proposal we will investigate how helminths directly promote cellular plasticity to enhance mucosal healing responses on a single cell level. We anticipate that these studies will uncover previously unrecognised mechanisms of tissue repair and regulation of the immune response.

Keywords: single cell RNA-sequencing, stem cells, regeneration, intestine, parasite, infection and immunity

PI: Irah King (Associate Professor)

Department: Microbiology and Immunology

email: irah.king [at]


Project title: Mice PainTracker.AI

Development of an AI-driven software to automatically track rodent pain behaviors and also detect changes in environmental complexity to implement the appropriate corrections to the detection parameters.

Keywords: Tracker, freely moving mice, segmentation, behavior quantification

PI: Hugues PETITJEAN (Research Associate - Data manager)

Department: Physiology

email: hugues.petitjean [at]


Project title: Genetic and epigenetic variability in the response of alveolar macrophages to M. tuberculosis

Studies in the Schurr lab are focused on understanding host genomic responses in tuberculosis (TB) and leprosy, the two most common human mycobacterial infectious diseases. TB is the main cause of mortality among people living with HIV. The vast majority of TB deaths occur shortly after infection with Mycobacterium tuberculosis (Mtb) pointing to the need of preventing infection of persons exposed to Mtb. Existence of host resistance to Mtb infection is supported by multiple lines of evidence. Understanding the molecular basis of infection-resistance will be critical to decrease the burden of TB in general and specifically among HIV infected people. The central hypothesis evaluated in our lab is that alveolar macrophages (AMs), the first pulmonary cells that encounter inhaled Mtb bacilli, are important effectors of infection-resistance. Hence, we aim to identify inter-individual genetic and epigenetic differences in the response of AMs to Mtb for HIV infected and healthy subjects.

Keywords:  epigenetic, HIV infection, infection and immunity, single cell RNA-sequencing, high-dimensional data, tuberculosis

PI: Erwin Schurr and Vinicius Fava (James McGill Professor)

Department: Human Genetics

email: vinicius.medeirosfava [at]


Project title: Elucidating the role of the MNK1/2-eIF4E axis on the immune landscape of postpartum breast cancer

Postpartum breast cancer (PPBC) is a type of breast cancer diagnosed within 5 years of having had a last child. Women diagnosed with PPBC have an increased risk of metastasis when compared to women with nulliparous breast cancers. We have been studying the mechanisms that contribute to postpartum breast cancer being so highly metastatic. In an international collaboration, we constructed a tumor microarray (TMA) containing tumors from women diagnosed with breast cancer during pregnancy, during the postpartum period, or outside of those reproductive stages. The TMA was stained with a panel of antibodies to characterize the immune landscape of the tumors, and was subjected to Imaging Mass Cytometry (IMC). IMC is a state-of-the-art technology that combines time of flight mass cytometry CyTOF systems and metal-conjugated antibodies with histology, allowing for the simultaneous detection of up to 30 proteins on a single cell level. Included in our IMC profiling are proteins known to play an important role in promoting metastasis, known as MNK1 and eIF4E. We seek to understand which cells within the tumor bed, such as tumor cells, fibroblasts, immune cells, endothelial cells, express most activated MNK1 and eIF4E. The latter requires analysis of the IMC data on a single cell level.

Keywords: Imaging mass cytometry, immunology, breast cancer

PI: Sonia del Rincon (Assistant Professor)

Department: Experimental Medicine

email: christophe.goncalves [at]


Project title: Is skeletal mechanotransduction similar in mammals and fish?

Bone is a tissue that continually adapts to changing external loading conditions via (re)modeling (modeling and remodeling) processes. Bone (re)modeling is accomplished by bone-resorbing osteoclasts, bone-forming osteoblasts, and osteocytes, terminally differentiated osteoblasts enwalled in the bone matrix. Osteocytes are thought to sense load applied to the skeleton. Surprisingly, the most evolutionarily advanced fish entirely lack osteocytes yet they remain responsive to mechanical loading. We have RNA-seq and microarray data from experiments in fish with (zebrafish) and without osteocytes (medaka) as well as in mice (C57Bl6J and Balbc) that involved swim training and tibial loading, respectively. We will study how the different organisms respond mechanical loading at the genetic level. Our main objective is to provide insights into the molecular mechanism(s) underlying bone mechano-transduction. Understanding mechanotranductive signaling pathways shared and unique to different species (fish and mice) may help in the development of novel bone-forming treatments for patients with skeletal diseases. 

Keywords: bioinformatics, mechanobiology, comparative biology

PI: Bettina Willie (Associate Professor)

Department: Pediatric Surgery

email: bwillie [at]


Project title: Bioinformatic pipeline development for the stratification of colorectal cancer liver metastasis

Colorectal cancer is the second leading cause of cancer in Canadians with liver metastases being the major cause of death. Three distinct histological growth pattern of colon cancer liver metastases have been described that affect patients' prognosis. The goal of this project is to use bioinformatic-based tools to generate a multi-modality signature-based on RNA sequencing data form resected tumor and background liver human specimens, to identify Colorectal Cancer Liver Metastases (CRCLM) patients who will 1) respond to Angiogenic Inhibitor-based therapies and 2) monitor the development of drug resistance (non-responders). It is expected that this knowledge will lead to optimization of current treatment strategies, a precision therapy approach to the management of metastatic disease and cost savings for the Canadian health care system.

Keywords: RNAseq, biomarkers , precision medicine, cancer, bioinformatics, risk stratification

PI: Anthoula Lazaris (Scientist)

Department: Surgery

email: anthoula.lazaris [at]


Project title: Mechanism of Radiation-induced damage in tissues and its control by Mesenchymal stem cell secretome 

Radiation therapy remains the first line of treatment for carcinomas, which inherently leads to radiation damage in surrounding normal tissues. No standard of treatment is available for the control of this damage. The secretome of mesenchymal stem cells has been seen to improve salivary gland cell function in irradiated mice model which is one of the most sensitive organs.

Radiation-induced injury is complex and not well understood. We hypothesize that a systems analysis approach can serve as a powerful tool to elucidate this complex pathophysiology. To the best of our knowledge, studies have been seen to identify a few molecular targets but there is less clarification on the pathway and systems biology to control this damage.  We aim to integrate the current data available for differential transcriptomics in radiated tissues to do a meta-analysis of the transcriptomics database. Gene enrichment and pathway analysis will be done to build a network and thus identify Gene ontology terms highly associated with radiation damage; if possible, identified pathways unique to normal cells so that the cancer cells are not targeted or are less targeted.

Secretome derived from Mesenchymal stem cells is a mixture of a multitude of proteins. Enrichment and pathway analysis for the datasets of the secretome from Mesenchymal stem cells will be performed to narrow down the proteins responsible for inducing possible changes and provide a predictive overview of the ongoing process. Eventually, we expect to develop/predict possible treatment for radiation damage. It will also serve as a good resource for researchers working in the areas of radiation biology, radiogenomics, and regenerative medicine.

Keywords: complex systems, data mining, genomics, precision medicine, radiology, AI, artificial intelligence, regenerative biomaterials, Big data analysis, Complex systems modeling, radiomics, Protein structure, uk biobank, regeneration, Human genetics, bioinformatics, cancer

PI: Simon Tran (Graduate Program Director)

Department: Faculty of Dentistry

email: simon.tran [at]


Project title: AI applied to FTIR images of skeletal tissues

Musculoskeletal tissues, such as bone and cartilage, resist deformation and fracture through their multi-scale structure. Aging and disease can lead to changes in composition and structure that compromise mechanical integrity. Therefore, assessments of tissue quality are integral to our understanding of the effects of disease on the musculoskeletal system. Fourier transform infrared (FTIR) spectroscopy is a powerful technique to investigate tissue composition. Thousands of spectra can be acquired across a cross-section of tissue to map compositional parameters, such as the mineral-to-matrix ratio, crystallinity, and collagen maturity. We have FTIR data from a cohort of pediatric human bone samples to investigate bone quality during longitudinal bone growth and from human cartilage samples to investigate changes in cartilage composition in osteoarthritis. This collaboration would work towards developing AI algorithms that provide deeper insight into musculoskeletal tissue quality.

Keywords: Tissue quality, Bone, FTIR spectroscopy, AI, Cartilage

PI: Elizabeth Zimmermann (Assistant Professor)

Department: Faculty of Dentistry

email: ezimmermann [at]


Project title: Bioinformatic analysis of immune infiltration of colorectal cancer liver metastasis

Colorectal cancer is the second leading cause of cancer in Canadians with liver metastases being the major cause of death from this disease. Two main histological growth pattern (HGP) of colon cancer liver metastases have been described. The first HGP derives its blood supply from angiogenesis and has a better prognosis while the second employs blood vessel co-option and has a poorer outcome. We have identified major differences in the number and types of immune cells infiltrating these two growth patterns. Using RNAseq, protein spatial profiling and bioinformatics analysis, we will establish a complete immune profile of colorectal cancer liver metastases and study its effect on tumor progression.  The immune system plays a primordial role in the destruction of tumors cells and many new therapies aim to increase its efficiency. It is expected that this knowledge will allow for the development of immunotherapies to use in combination with current treatment

Keywords: RNAseq, bioinformatics, metastasis, protein spatial profiling, cancer, immunology

PI: Peter Metrakos (Professor)

Department: Surgery

email: peter.metrakos [at]


Project title: Analysis of microRNA, gene expression and single-cell RNAseq data in squamous cell carcinoma

Head and neck squamous cell carcinoma (HNSCC), is one of the most prevalent types of malignancy and its incidence is increasing globally. Despite intensive research, there has been limited success in blocking recurrence and metastasis that occurs in a sizable proportion of HNSCC patients. Transforming growth factor-beta (TGF-beta) is a multifunctional growth factor that exerts potent pro-metastatic effects in cancers, including HNSCC. Thus, blocking its pro-metastatic effects represents a promising strategy for the treatment of HNSCC. We have identified CD109, a cell surface GPI-anchored protein, as a potent antagonist of TGF-beta signaling and responses in vitro and in vivo. Emerging evidence indicates that CD109 expression is dysregulated in a large number of HNSCC patients, and that CD109 expression positively correlating with cancer-specific survival. However, CD109 function in HNSCC has not yet been determined.  We have recently performed gene expression profiling and miRNAseq analysis on SSC cells with CRISPR-mediated CD109 knockout (KO) SCC (A431) and parental cells and have identified differentially expressed genes (DEGs) and miRNAs (DEMs) in these samples. Preliminary analysis of our mRNA profiling data suggests that CD109 inhibits epithelial-mesenchymal transition (EMT) and regulates multiple intracellular pathways in A431 cells. Moreover, our miRNA profiling data indicates that CD109 regulates the expression and function of specific miRNAs implicated in EMT and HNSCC. Our future goal is to perform single-cell RNA sequencing (scRNAseq) on xenograft tumors derived from CD109 KO and parental cells and on HNSCC patient tumor samples to understand the role of CD109 in tumor progression and cancer cell heterogeneity. Together these findings provide compelling reasons to understand the molecular mechanisms by which CD109 regulates TGF-beta-mediated pro-metastatic effects in HNSCC, setting the stage for the proposed research.

The proposed research will combine the experience and expertise of a basic scientist (A Philip) and a computational biologist to integrate miRNAseq, gene expression profiling and scRNA-seq data for the potential discovery of new biomarkers or molecular targets for the treatment of HNSCC. We expect that the collaboration will allow us to generate unique results which we will publish in a high impact journal in co-authorship with the computational biologist. In addition, results from the proposed research may form the basis of a new hypothesis which can be put forth in future grant applications. Also, the proposed research may provide an opportunity for a new graduate to pursue an MSc degree under the co-supervision of A Philip and the computational biologist.

Keywords: Big data analysis, bioinformatics, CD109, squamous cell carcinoma, high-dimensional data, data mining, miRNA-seq, single cell RNA-sequencing

PI: Anie Philip (Professor)

Department: Surgery

email: anie.philip [at]


Project title: Exploring the intratumoral heterogeneity of prostate cancer and its potential clinical implications

Prostate cancer (PCa) is the most commonly diagnosed visceral malignancy in North American men and a leading cause of cancer-related mortality. A key issue in the clinical management of PCa is the difficulty to accurately distinguish aggressive from indolent tumors, which inevitably leads to overtreatment of harmless diseases and undertreatment of aggressive cancers with metastatic potential. We first have identified biologically and clinically relevant PCa molecular subtypes based on distinct gene expression patterns. These expression subtypes exhibited distinct patterns of DNA copy number alterations (CNAs) supporting the existence of alternative parallel pathways of tumorigenesis. In addition to this molecular heterogeneity observed among tumors, a growing body of evidence suggests that there are molecular variations even within each tumor. The biological and clinical implications of such diversity remain to be determined in PCa. We hypothesize that the molecular intratumoral heterogeneity may impact the natural history of PCa and treatment selection. The goal of this study is to assess the molecular intratumoral heterogeneity and its potential consequences on tumor classification and outcome. This goal will be achieved by analysing the gene expression and CNA data derived from RNA and DNA extracted from two distinct areas of tumors collected from a cohort of PCa patients treated by radical prostatectomy with clinicopathologic correlates. This project should shed light on the biology of PCa progression and ultimately lead to novel strategies of biomarker and targeted therapy development.

Keywords: prostate cancer, biomarkers , genomics, DNA copy number alteration, gene expression

PI: Jacques Lapointe (Associate Professor)

Department: Department of Surgery, Division of Urology

email: jacques.lapointe [at]


Project title: Determining mechanisms of ILC2-mediated CD4 T cell priming during allergic lung disease.

The type 2 immune response is critical for host defense against large parasites such as helminths, wound healing and body metabolism. On the other hand, dysregulation of type 2 immunity causes immunopathological conditions, including asthma, atopic dermatitis, rhinitis, and anaphylaxis, which have been referred to as type 2 immunopathologies. Thus, a balanced type 2 immune response must be achieved to mount effective protection against invading pathogens while avoiding immunopathology. The proposed research aims to decipher the mechanisms between innate and adaptive immune regulation at the single cell level and will as such give detailed insights into the molecular regulation of type 2 immunopathologies.

Keywords: Asthma, RNAseq, single cell RNA-sequencing, Allergic lung disease, group 2 innate lymphoid cells (ILC2), CD4 T cells

PI: Jorg Fritz (Associate Professor)

Department: Microbiology & Immunology

email: jorg.fritz [at]


Project title: Modelization of light transmission to photoreceptor cells in the mouse retina

Light entering the eye has to travel through several layers of retinal neurons before reaching the light-detecting photoreceptor (PR) cells located at the back of the retina. Only recently was this conundrum partially explained by the capacity of Müller cells, a specialized glial cell, to guide light directly from the front to the back of the retina, acting as a sort of “optic fiber”. However, how light is transmitted at the Müller/PR cells interface still remains unknown. To tackle this question, we recently reconstructed the ultrastructure of the mouse retina in 3 dimensions using the focused-ion beam serial electron microscopy technology. Three independent acquisitions were generated with each set comprising on average 850 images for 2Go. The exhaustive spatial reconstruction of the numerous retinal structures for each set, that can only by achieved using the power of artificial intelligence-based image segmentation, will allow to precisely model light transmission in the retina. We anticipate this work to uncover structural basis important for visual acuity since preliminary results obtained in our laboratory show that Müller cells spatially couple with individual cones, a type of PR specialized in colour discrimination and high acuity vision that specifically degenerate in a number of retinal diseases causing vision loss like macular degeneration and cone-rod dystrophy.

Keywords: Modelization, FIB-SEM, Retina, Image Segmentation, Vision

PI: Michael Housset (Post Doc)

Department: Cellular Neurobiology Research Unit

email: michael.housset [at]


Project title: Quantitative evaluation of kidney aging 

The functions of human organs deteriorate during aging. The most severe functional decline occurs for the kidney. Moreover, the kidneys of elderly individuals are particularly vulnerable to acute kidney injury, which often progresses to kidney failure. As a result of this aging-related deterioration, more than 50% of the elderly suffer from kidney dysfunction. Ultimately, kidney malfunction triggers the decline of overall health, because it compromises the performance of other organs. As the Canadian population ages, there is an urgent need for new strategies to preserve kidney function.

The routes to cellular senescence are often organ-specific. Our laboratory focuses specifically on kidney aging and senescence. In the context of the MiCM Research Match we have two Specific Aims:
(1) The changes in the kidney transcriptome that are related to cell aging and senescence.
(2) The rewiring of cell signaling pathways in aging and senescent kidney cells.

Long-term Goals. We believe that our study will (a) short-term generate new information on basic physiological processes as they pertain to the kidney, and (b) long-term, identify new strategies to protect kidney health in aging individuals. A collaborative and multidisciplinary approach is mandatory to achieve these goals.

Keywords: Cellular senescence; aging biomarkers, Complex systems modeling, cell signalling, kidney health, transcriptome analysis, imaging

PI: Ursula Stochaj (Associate Professor)

Department: Physiology

email: ursula.stochaj [at]



Population Health Research

Project title: Integrating machine learning applied to risk prediction of stress-related mental disorders

The overarching goal of this research program is to develop risk prediction models using integrative and innovative computational methods to better identify and predict the onset of stress-related mental disorders (SMD) and their associated health utilization patterns. This program builds upon two closely connected CIHR-funded programs- Zone d’Épidémiologie Psychiatrique du Sud-Ouest de Montréal (ZEPSOM) Epidemiological Catchment Area Study and ZEPSOM-Bio (Genetic and Epigenetic study of ZEPSOM cohort). By merging the existing data, including genomic, epigenetic, longitudinal epidemiological surveys, and linked Quebec health administrative data, we will: determine psychosocial predictors associated with SMD onset and severity; establish the risk predictive models for SMD onset and severity; examine the relationship between risk predictive models and mental health utilizations; explore vulnerability and resilience among those with the history of childhood maltreatment; and, conduct a pilot study to test the the potential use of risk predictive models of SMD in a clinical setting.

Keywords: machine learning, genomics, personalized treatment, population, data mining

PI: Xiangfei Meng (Assistant Professor)

Department: Psychiatry

email: xiangfei.meng [at]


Project title: Who gets coronary revascularization?

This project proposes to explore the patient, physician, institutional and regional determinants to receiving advanced cardiac treatments. Using big data and sophisticated analytical techniques, the project aims to dissect the relative contributions of these factors in determining accessibility to coronary revascularization procedures. 

Keywords: evaluative research, data mining, disease cohorts, statistical analyses, Big data analysis, myocardial infarction 

PI: Jay Brophy (Professor)

Department: Medicine

email: james.brophy [at]


Project title: Large-scale Recurrent Disease Progression Networks for Modelling Risk Trajectory of Heart Failure

Recurrent neural networks (RNN) are powerful frameworks to model medical time series records. Recent studies showed improved accuracy of predicting future medical events (e.g., readmission, mortality) by leveraging large amount of high-dimensional data. However, very few studies have explored the ability of RNN in predicting long-term trajectories of some acute onsets due to the disease complications such as heart failure hospitalizations among patients with congenital heart disease. This is mainly because of the lack of an efficient method that can model long-term medical history while mining high-dimensional healthcare information. Also, interpretability of the deep learning models is often in question as whether they can provide clinician's understandable medical terms for the follow-up investigation. We propose a highly scalable and interpretable deep recurrent topic (DRT) model to address the above challenges. To demonstrate its real-world utility, we will apply DRT to Quebec claim data for 80,000 patients with 20 year medical follow-up.

PI: Yue Li (Assistant Professor)

Department: School of Computer Science

email: [at]


Project title: Is an Image Worth a Thousand Words:  from Spine radiographs to Quality of Life and Mobility?

Back pain is a common problem among adults. There is a lack of information about the effectiveness of imaging to appropriately identify adults with worse symptomatology or worse mobility leading to confusion and frustration on the part of patients.

Using the radiograph images and data collected for the Canadian Multicentre Osteoporosis study (CaMos;; 1995- 2018) in over 6000 men and women  50 years and older who have undergone spinal radiographs at baseline and every 5 years for 15 years, we propose to develop algorithms that would associate spine radiograph images with quality of life (including pain) and mobility outcomes in older adults.

Keywords: pain, radiology, population, Image analysis and machine learning, quality of life, mobility

PI: Suzanne Morin (Associate Professor)

Department: Medicine

email: suzanne.morin [at]


Project title: Prediction of Aortic Stenosis using Artificial Intelligence 

Aortic stenosis (AS) is a serious cardiovascular disease characterized by reduced blood flow due to a constriction of the aortic valve. Genome-wide association studies (GWAS) performed by our group and others have identified genetic variants with significant associations with AS. Artificial intelligence methods may increase the predictive power of genetic variation.  While each SNP from a GWAS may not have a large effect size, SNPs can be combined to capture a greater proportion of genetic variance. This approach generates a genetic or polygenic risk score (PRS) and can be used to assess the additive effects of multiple variants leading to improved prediction.  These models can be improved by the inclusion of other risk factors in the model.  However, artificial intelligence methods may further increase the predictive power of genetic variation.  Clinical and genetic data obtained from a European-ancestry case-control dataset from the Genetic Epidemiology Research on Aging (GERA) study will be used to train and test neural network (NN) models using either clinical data only, genetic data only, or combined clinical and genetic data. Importantly, these models will then be validated with AS cases from the independent UK Biobank dataset.

Keywords: artificial intelligence, GWAS, clinical prevention, cardiovascular disease, uk biobank, aortic stenosis, machine learning

PI: George Thanassoulis (Associate Professor)

Department: Medicine

email: george.thanassoulis [at]


Project title: Artificial intelligence informed referral to tertiary care for congenital heart disease patients

Increased survivorship in patients with congenital heart disease results in longer time windows to express important clinical sequelae of heart failure. Such patients require careful surveillance with expert care. Even in Canada where health insurance is universal, less than 30% of adult with congenital heart disease are being systematically followed in specialized tertiary care centers, although referral to such centers reduces mortality rate. In this study, we propose to use artificial intelligence methodologies to better direct referral to tertiary care center. We will create and deploy risk calculators for identifying high risk patients and informing the timing of referral. Working in collaboration with data scientists we aim to realize precision delivery of health services to improve severe cardiovascular outcomes.

Keywords: referral, data mining, precision medicine, quantitative medicine, personalized treatment, chronic disease, congenital heart disease

PI: Ariane Marelli (Professor)

Department: Medicine

email: ariane.marelli [at]


Project title: Transience of Liver Transaminase Increases During Latent Tuberculosis Treatment

One of the greatest barriers to provision of latent tuberculosis infection treatment is the high-risk for adverse events with current regimens--particularly hepatotoxicity. Current guidelines for managing hepatotoxicity encourage continuation of treatment during mild elevation of liver transaminases, since these are often transient. What is uncertain, however, is if there are patient, clinical, or other laboratory parameters that might increase one's risk for continued elevation of liver transaminases. This would permit targeted pause or change of therapy for patients at high-risk of continued elevation, preventing patient morbidity. This project would utilize detailed clinical trial data of >6000 patients treated with two of the most common regimens for latent tuberculosis infection with the aim of identifying if there are factors that would permit risk stratification.

Keywords: risk stratification, liver toxicity, tuberculosis

PI: Jonathon Campbell (Post Doc)

Department: Epi, Biostats, Occupational Health

email: jonathon.campbell [at]


Project title: Communicating Risks and Benefits: Updating the Online Tuberculosis Risk Calculator "TSTin3D"

TSTin3D ( is an online tool that is used by clinicians to help determine cumulative risk of tuberculosis among people who have a positive diagnostic test for latent tuberculosis. To further help decisions on whether to provide latent tuberculosis treatment, it also communicates risk of hepatotoxicity with isoniazid, a latent tuberculosis treatment option. Since its creation, much more is known on risk of tuberculosis among people with positive diagnostic tests and various health conditions. As well, newer regimens for latent tuberculosis, such as rifamycin-based regimens, are available and risks of hepatotoxicity (and other adverse events) are not currently communicated on the website. To support shared decision-making between patients and providers, we wish to use data we have collected on tuberculosis risk among people with various health conditions and safety data on newer rifamycin-based treatment regimens to update TSTin3D, which will require both data analysis and web-based programming.

Keywords: communication, risk prediction, clinical prediction, tuberculosis

PI: Jonathon Campbell (Post Doc)

Department: Epi, Biostats, Occupational Health

email: jonathon.campbell [at]


Project title: Sex and gender-related factors in chronic noncommunicable diseases 

Background: GOING-FWD (Gender Outcomes INternational Group: to Further Well-being Development), is a personalized medicine project recently funded by the Canadian Institutes of Health and Research and GENDER-NET+, and is part of the European EU H2020 initiative. Psycho-cultural-social factors, which are associated with the gender of individuals, are rarely collected as determinants of health outcomes, despite their powerful contribution to overall well-being. Gender-related factors including gender identity, gender roles, gender relations and institutionalized gender are assumed to differ between sexes. Objective: As a pilot study for the GOING-FWD consortium, the aim for this proposal is to develop a harmonized subdataset that will be used for conduct a federated analysis of health related outcomes based on clinical and gender-related factors known to be associated with patient sex.

Keywords: quantitative medicine, disease cohorts, precision medicine, statistical analyses, chronic disease

PI: Louise Pilote (Professor)

Department: Medicine

email: louise.pilote [at]


Project title: Influence of data processing on the identification of contaminants in food and human breast milk. 

As part of the CIHR funded team project “Endocrine Disrupting Chemicals: Towards Responsible Replacement” (PI: Prof B. Hales, McGill), food (n=504) and human breast milk (n=200, MUHC) samples have been collected in the Montreal region to screen for the presence of historical and emerging contaminants ( Food samples have been analyzed using HPLC-HRMS and this has already resulted in >300 gigabits of raw data for food samples. Similar amount of data is being processed at the moment for human breast milk. The objective of the proposed study is to compare the data processing of these data using different algorithms, and identify which parameters are critical for the detection and identification of trace contaminants. This work will allow for a better interpretation of metabolomics ‘big data’ to map the xenobiotics in these samples. The project is proposed in collaboration with Drs Bayen, Hales and Goodyer. 

Keywords: Non-targeted Analysis, Human Breast Milk, Exposome, Chemical Exposure, Mass Spectrometry

PI: Stephane Bayen (Assistant Professor)

Department: Department of Food Science

email: stephane.bayen [at]



Back to top