Students

Introduction: Chromosomal disorders such as trisomy 21 (most common), 18 and 13 are a source of concern for parents, in terms of fetal health, delivery/miscarriage; this may be compounded by concerns about financial issues. These concerns increase with the age of the mother (1,2). Prenatal screening is used to assess the likelihood of having a fetus with such anomaly(ies) and if necessary further diagnostic tests. Previous studies have shown that pregnant women actively seek information to make informed decisions about testing: what options are available? Are they prepared enough to make a choice? Do they have all the necessary information? How can their own values and preferences be considered?

Effective decision support tools exist to help people facing difficult decisions to make informed choices (4). However, the evidence contained in these tools has not been evaluated based on the relative weight it contributes to the decision to be made. Furthermore, the intentions of pregnant women and their partners to use such a tool in a digital format are unknown (5).

Objective: The main objective of our study is to assess the intention of pregnant women and/or their partners to use a mobile application-based decision support tool for prenatal screening. More specifically, it will: identify potential factors that may influence decision making; assess the relative weight of the information contained in the tool ascribed by pregnant women and/or their partners; and assess their intention to use it.

Methods: Study design: This is a descriptive cross-sectional study that represents phase 1 of a project called APP4WE (Analytical mobile application to support shared decision making for pregnant women) of the Canada Research Chair in Shared Decision Making and Knowledge Translation (6). This project aims to enable pregnant women and their partners to get the support they need to make informed decisions about prenatal screening and includes several phases. Participants and sample size: For phase 1, we propose to recruit an independent sample of 90 pregnant women and their partners from three clinical sites (a midwife-led birthing centre, a family practice clinic, and an obstetrician-led hospital clinic) in Quebec City and Montreal, Canada. Pregnant women and their partners will be recruited to reflect the respective proportions of socio-economic, ethnic, and linguistic communities. To be eligible for the study, pregnant women must be at least 18 years old, more than 20 weeks pregnant, have a low-risk pregnancy, not have given birth near the dates of data collection, be able to speak and write French or English and be able to give informed consent. Partners of pregnant women will also be asked to provide informed consent. Measured outcome: The primary outcome of this study is to measure the intention of pregnant women and/or their partners to use a mobile application for prenatal screening decision making. To assess this outcome, the Continuing Professional Development - Feedback (CPD-Feedback) questionnaire was used. This tool is a validated 12-item questionnaire that assesses the impact of continuing professional development activities on the clinical behavioural intentions of health professionals. We will also determine the factors (socio-demographic and others) potentially associated with intention. Statistical analysis: We will first perform descriptive statistics to determine the characteristics of our study population and the distribution of intention. Subsequently, a linear mixed model will be used to determine potential factors influencing the intention of pregnant women and/or their partners to use a mobile application for prenatal screening decision making. We will specify random effects at the practice level (cluster), which will allow us to answer our research question while considering the hierarchical structure of the study.

Références :
1. Ohman SG, Grunewald C, Waldenström U. Women's worries during pregnancy: testing the Cambridge Worry Scale on 200 Swedish women. Scand J Caring Sci. 2003 Jun;17(2):148-52. doi: 10.1046/j.1471-6712.2003.00095.x. PMID: 12753515 

2. Practice Bulletin No. 163: Screening for Fetal Aneuploidy. Obstet Gynecol. 2016 May;127(5): e123-e137. doi: 10.1097/AOG.0000000000001406. PMID: 26938574. 

3. Légaré F, St-Jacques S, Gagnon S, Njoya M, Brisson M, Frémont P, Rousseau F. Prenatal screening for Down syndrome: a survey of willing in women and family physicians to engage in shared decision-making. Prenat Diagn. 2011 Avr;31(4):319-26. doi: 10.1002/.2624. EPUB 2011 Jan 26. PMID : 21268046.

4. Agbadje TT, Pilon C, Bérubé P, Forest JC, Rousseau F, Rahimi SA, Giguère Y, Légaré F. User Experience of a Computer-Based Decision Aid for Prenatal Trisomy Screening: Mixed Methods Explanatory Study. JMIR Pediatr Parent. 6 septembre 2022;5(3):e35381. doi : 10.2196/35381. PMID : 35896164; Numéro PMCID : PMC9490528.

5. Delanoë A, Lépine J, Turcotte S, Leiva Portocarrero ME, Robitaille H, Giguère AM, Wilson BJ, Witteman HO, Lévesque I, Guillaumie L, Légaré F. Rôle des facteurs psychosociaux et de la littératie en santé dans l’intention des femmes enceintes d’utiliser un outil d’aide à la décision pour le dépistage du syndrome de Down : un sondage en ligne fondé sur la théorie. 2016 Oct 28;18(10):e283. doi : 10.2196/jmir.6362. PMID : 27793792; PMCID : PMC5106559.

6. Abbasgholizadeh Rahimi S, Lépine J, Croteau J, Robitaille H, Giguere AM, Wilson BJ, Rousseau F, Lévesque I, Légaré F. Facteurs psychosociaux de l’intention des professionnels de la santé d’utiliser un outil d’aide à la décision pour le dépistage du syndrome de Down : étude quantitative transversale. 2018 Apr 25;20(4):e114. doi : 10.2196/jmir.9036. PMID : 29695369; PMCID : PMC5943629.

Duel-energy Computed Tomography (CT) imaging has the potential to better characterize materials. DE CT images would allow for a more accurate identification of tissues present in the human anatomy. The presence of highdensity elements (e.g. region of the shoulder, posterior fossa, metallic inserts, etc.) in the scanned subject causes deterioration of the CT image quality (e.g. beam-hardening artifacts). The polychromatic nature of the X-ray beam used in CT scanners is the origin of some image artifacts. In this work, we propose a physics-rich polychromatic projection model that uses the spectrum information, the detector response, the filter geometry and a calibration curve. This model is embedded in an iterative reconstruction algorithm, and inherently reduces beam-hardening artifacts. With dual-energy acquisitions, one can reconstruct quantitative images, with effective atomic number, and electron density information. Besides that, various reconstructions techniques are explored, so high-quality images can be obtained with less artifacts, ultimately, improving the characterization and identification of elements in the image.

Prostate cancer is the second most frequent cancer and the fifth leading cause of cancer death among men. To improve patient outcomes, treatment must be personalized based on accurate prognosis. Nomograms already exist to identify patients at low risk for recurrence based on preoperative clinical information, but these tools do not use patients’ medical images.

The goal of this project is to use deep learning to develop a model combining FDG-PET/CT images and patient clinical data to improve the pre-treatment prognosis of high-grade prostate cancer. This model must be efficient, but also interpretable in order to allow an expert to understand the given probabilities.

Mariame Gnéré Coulibaly's project will focus on the development of a tool for the prediction of bacterial host of phages by machine learning (ML). The first objective will be to develop a ML tool to identify bacteria-phage pairs from CRISPR spacer sequences (clustered regularly interspaced short palindromic repeats) found in bacterial genomes, which are fragments of phage genomes that have infected the bacteria. The second objective is to develop a ML tool to identify bacteria-phage pairs based on sequence and methylation information (i.e. the addition of methyl groups to nucleotides). For the first two objectives, algorithms such as neural networks with attention mechanisms, and similarity predictors based on string kernels, will be developed and tested.

The third objective is to develop a multi-view algorithm combining the two previous objectives, with one view for CRISPR information and a second view for methylation patterns.

The tools developed will have a strong impact on the microbiology and virology research community, by being able to identify new bacteria-phage pairs from microbiota samples.

Fadwa Mehdaoui's project focuses on the analysis of interactions between bacteria and their viruses, called phages, in the microbiota using metagenomic data coupled with bioinformatics and machine learning methods.

The metagenomic data come from the North Sentinel project 3.6 (axis Environment-health interactions in the North) that sampled the gut microbiota of young Inuit from Nunavik. Following the identification of bacteria and phages in the sequencing data, her work will consist of an exploratory statistical analysis of the interactions between bacteria and phages, followed by a machine learning analysis based on interpretable algorithms such as the set covering machine or random forests. It is also envisaged to develop a new machine learning model for phage host prediction based on these data.

This project could have important implications for the understanding of the interactions between bacteria and phages, which are very poorly known, but also for the knowledge of the gut microbiota of the Inuit in relation to their unique diet.

Prediction and early identification of stroke is crucial to prevent emergency department (ED) revisits and initiate treatment, reducing morbidity and mortality.

This project focuses on the analysis of non-contrast brain CT (NCCT) data to predict early ED revisits for patients coming back with a stroke diagnosis. The first objective will be gathering open-source NCCT data as well as NCCT data from the Integrated Health and Social Services Center from Chaudiere-Appalaches (CISSS-CA) to classify the presence/absence of stroke using an existing model. The second objective will be to develop and test a machine learning model with weights from the previous model and other relevant clinical data to classify short-term revisits to the ED as an outcome.

From a clinical perspective, the development of such a tool may help support neuroradiologists in image interpretation and clinical decision making in the ED. 

Delirium is a condition that, when left unmanaged, is associated with increased mortality and longer hospitalization of patients in intensive care; therefore, its detection should be an integral part of care. It is characterized by confusion, anxiety and reduced alertness. It is estimated that 75% of delirium cases are not detected on admission to hospital. Detecting such an acute condition requires frequent monitoring of participants, which is labor intensive and requires expertise. However, the participants' vital signs, which can be collected continuously throughout their stay in intensive care, could contain information indicative of the present state of consciousness, and possibly predictive of the future state.
Our goal is to build an automatic machine learning classifier based on vital sign data to (a) identify times when the patient was delirious, and (b) predict delirium incipience. As a primary measure, we will use a clinically validated tool, the Confusion Assessment Method for Intensive Care Unit (CAM-ICU). This assessment was performed twice a day, once in the morning and once in the afternoon, in our study population at the CISSS de Chaudière-Appalache (Hotel Dieu de Lévis). The learning algorithm will be trained on the participants' vital signs before, after, and during the delirium episodes in order to (a) extract the vital sign characteristics related to a delirium state; (b) the probability that the patient is delirious or not, based on these characteristics; and (c) the probability that the patient will develop a delirium state within a reasonable time window (e.g. 1 hour).
Even if the machine learning model does not reach the accuracy and precision of a validated questionnaire, its use in healthcare facilities would optimize care, mainly by drawing attention to any suspicious drift (high sensitivity). Considering that patients who remain with untreated delirium are associated with higher mortality rates and longer ICU stays, a clinical indicator such as this model can help the care team manage this otherwise unnoticed symptom.
 

The risk associated with the stochastic effects of neutron radiation is known to be strongly energy dependent. Over the past decade, several studies have used Monte Carlo simulations to estimate the relative biological effectiveness (RBE) of neutrons for various types of DNA damage in order to understand its energy dependence at the fundamental level. However, none of these studies implemented DNA repair simulations in their pipeline.

In this project, we investigated the effects of adding repair mechanisms to Monte Carlo-based RBE estimates of DNA damage by neutrons. Our group had previously carried out condensed history (CH) simulations to profile the energy spectrum and relative dose contribution of the secondary particles produced by neutron interactions in tissue. In this project, we use the results of our CH simulations to simulate the irradiation of TOPAS-nBio’s DNA model by a flat spectrum of neutrons ranging from 1 eV to 10 MeV, as well as reference X-rays at 250 keV. Induced DNA damage are recorded using the standard DNA damage data (SDD) format abd DNA repair are simulated using the DNA Mechanistic Repair Simulator (DaMaRiS) framework.

Patient portal is an emerging healthcare technology that has shown promising effects in enhancing patient care experience and promoting patient health outcomes. Opal, a digital patient-centred portal, is currently available to patients at the Cedars Cancer Center at McGill University Health Centre (MUHC), providing real-time access to personal health information (upcoming appointments, clinical notes, lab results, etc.) in conjunction with the disease- and treatment-specific education materials. This project aims to quantify the initial impact of the Opal patient portal on operational and health-economic outcomes at the Cedars Cancer Centre. We leverage patient-level data from the Opal patient portal and MUHC. The primary method is to use propensity score matching to construct a matched cohort that compares operational outcomes and resource utilization between Opal users to non-users.

The planned outcomes of interest include missed appointments, emergency room visits, hospitalization, medical record requests, and fertility clinic appointments

Cancer patients go through a lot during treatment. Many patients need support from other patients who know what they are going through because family and friends, no matter how supportive they try to be, don’t always understand their struggles.
The Opal Health Informatics Group is developing two programs for supporting cancer patients in the portal Opal: OncoBuddy and OncoConseil. OncoBuddy is a support system that matches cancer patients with volunteer patients (we call them buddies) based on a matching algorithm that considers criteria selected by the patients themselves. OncoConseil, on the other hand, matches patients with threads of information that might interest them, such as tips and tricks to get a smoother experience during the treatment.
This project focuses on finding selection criteria for building the matching algorithm. Using semi-structured interviews with cancer patients, we will extract a panel of selection criteria that are the most valued by patients, such as type of cancer, age and gender of the buddy, stage of cancer, etc. 
 

Cancer patient peer support is a beneficial tool for current patients and previous patients who can share lived experiences. However, current peer support is inefficient as it is conducted manually and relies on a coordinator to match patients based on a few known factors.

Therefore, this research study will examine ways to develop AI-powered matching algorithms that will more efficiently and effectively match cancer patients according to a wider and more complex set of factors than can be done manually. In this research project, we are designing and developing an AI-matching algorithm for the OncoBuddy/OncoConseil project and evaluating its effectiveness to ensure recommended matches will result in appropriate peer support. We are comparing multiple existing AI models, (i.e. the Deferred Acceptance Algorithm and the Genetic Algorithm) and testing the models on synthetic patient data that we have generated with statistical inferences from the existing Opal database and Statistics Canada. A fitness function derived from previous research will determine the efficacy of the matching algorithms.

As a result, we have generated a synthetic dataset of 1770 patients to be used for training and testing purposes, implemented multiple AI algorithms, and deployed a live dashboard prototype.

The cancer experience and the uncertainty surrounding it is anxiety provoking. One way in which the non-clinical uncertainty of the cancer experience can be reduced is through peer support. The Opal Health Informatics Group seeks to evaluate the efficacy of an artificial intelligence-based peer support matching algorithm in the pre-existing patient portal Opal in the hopes of facilitating peer support programs (for cancer patients and their caregivers) in Quebec.

Our work focused on the architectural design, development, and demonstration of a proof of concept stand-alone web application used to demonstrate the results of several varieties of AI-powered matching algorithms on test patient data.

Supervised classification allows to build predictive models based on complex data to help human decision making processes. It has undergone an impressive development in recent years, particularly thanks to neural networks and the use of big data. However, these methods are not relevant to use on databases in which only a few instances are available to build the model, and even less when these instances are described by a large number of features. This type of problem, called fat data, is recurrent in the medical field, in which the extraction of data on patients is costly, but provides a large amount of information for each one. Moreover, in the medical field, it is common to perfrom several types of analysis on the same patient : genomic, metabolomic, transcriptomic, etc. This type of database is called multi-omics.

The goal of this project is to use and develop multi-view classification algorithms relevant to the processing of multi-omic fat data. 

In medical imaging, radiomic features make it possible to characterize heterogeneity of a region of interest at the anatomical level. This way of quantifying the heterogeneity of a region of interest can be useful, for example, in order to identify the more aggressive tumors in oncology. To do this, we hypothesize here that variation in magnetic resonance imaging (MRI) acquisition sequences and its resulting different levels of contrast would make it possible to optimize the subsequent radiomic analysis.
In this project, a pipeline for the analysis of real medical images will first be set up in order to quantify the robustness of radiomic characteristics according to variations in acquisition protocols. Then, an MRI acquisition simulation pipeline will be developed in order to evaluate the potential for optimizing radiomic features in medicine.
 

The PARTAGE project (Patients and Researchers Team-up and Generate Evidence) is a research project that is examining mechanisms to allow patients to securely share their clinical data with researchers using the Opal patient portal. As part of the overall PARTAGE project, this specific sub-project is about obtaining stakeholder feedback on the concept of data-sharing. To do so, we are using a process of stakeholder co-design in which patients, clinicians and researchers are embedded in the research team and we are obtaining additional feedback from each stakeholder group through focus groups and surveys.

In diagnostic radiology, the use of ionizing radiation is justified by benefits surpassing risks. From an epidemiological perspective, this balance is difficult to assess because accurate dose values for individuals are not available. This project consists in developing tools to automatically report dose-to-organs from Computed Tomography (CT) images. First, a machine-learning based, multiclass segmentation tool will be developed to automatically contour organs in CT imaging studies. Then, a fast GPU-based Monte Carlo code will be used to compute dose maps from technical scanning parameters stores in DICOM headers of medical images. A large database of dose-to-organ values will be constituted as well as interactive dashboards to explore dose usage as a function of site explored, device used, etc.

On the long term, this database will be linked with epidemiological cancer data to assess potential causal relations.

High-dose-rate (HDR) brachytherapy is a standard treatment modality to treat cancer (e.g., prostate and cervical cancer) using the ionizing radiation of a small encapsulated radioactive source. The curative aim in the clinic is to create treatment plans that maximize the dose to the tumor while minimizing the dose to normal tissues. When it comes to the treatment plan generation, manual fine tuning of an objective function is necessary to achieve optimal trade-offs between these two conflicting objectives. Therefore, the plan generation is a time-consuming iterative task for practitioners; the plan quality can be dependent on the user skills.

The purpose of the project is to implement efficient optimization algorithms on GPU that can generate thousands of alternative plans with optimal trade-offs (Pareto-optimal plans) within seconds. Using real-time plan navigation tools, the user can quickly explore the trade-offs through the set of Pareto-optimal plans and select the best plan for the patient at hand. The impact of these novel optimization approaches is quantified and compared to the standard clinical approach. 

The project consists in determining and exploring the possibilities offered by dynamic dashboards in a medical context as well as the associated data management structures. The project therefore considers several aspects of data management. In this sense, the considerations related to DICOM data transfers as well as different approaches to their management and conservation are considered. In addition, the dashboards will be designed to ensure an effective, clear and concise presentation with recognized visualization tools. Different additions will be made to the different portions of the project during its implementation depending on the direction taken by the research and the needs of health professionals. Particular emphasis is placed on compliance with FAIR principles by the resulting system.

It is often difficult to share denominated data between different organisations and researchers due to ethical constraints related to respondent's confidentiality. This is a frequent reality in healthcare, given the inherent sensitivity of the data involved. One option in this case is to not share the data directly, but rather to provide access to it via a tool that controls the risk of disclosure of the queries made and allows only those it considers safe. DataSHIELD is such a tool that has been proposed to protect the confidentiality of a dataset, and which is used via the statistical software R. It also allows statistical analyses to be carried out on several datasets hosted in different locations, always ensuring the confidentiality of the respondents. In this project, we are interested in the confidentiality guarantees provided by the software, and in its limitations. In particular, we study the potential uses of the software for advanced statistical analyses, such meta-analyses and the use of neural networks.

Treatment of cancer with radiation is a proven technique used worldwide. One of the ways to treat prostate cancer is by using brachytherapy either alone or as a boost. At the moment, the techniques used depend on the experience of the treatment team and researchers are trying to overcome this problem.
In our case, the technology considered to address this problem is deep learning. Therefore, the aim of this project is to use deep learning to develop tools for planning in high dose rate brachytherapy for the treatment of prostate cancer.
Four different phases are initially targeted. The first consists of a classification of treatment plans. The second is a “reinforce learning” approach to help optimize treatment plans, by modifying the optimization objectives in order to consider each patient in a unique way. The third is dose map prediction based on patient anatomy. The fourth is the generation of treatment plans; from the patient's anatomy or from a dose map to find an adequate treatment plan.
The proposed work is a new approach that will ultimately help with the planning of high dose rate brachytherapy treatments for prostate cancer.
 

Recent breakthroughs in medicine have shown a link between cerebrovascular pathologies and the risk of developing mild or major cognitive disorders of the like of Alzheimer’s disease (AD). The project will thus consist in developing a tool for the characterisation of the cerebrovascular system.
Indeed, the measurement of arteries and veins (diameter, density, etc.) in every brain region will
provide a better understanding of the vascular health changes throughout the aging process and if these changes are linked to two markers heavily correlated with neurocognitive disorders: cerebral tissue atrophy and cerebrovascular lesions (mainly hyperintensities in white matter and cerebral microbleeds). Thus, the cerebrovascular system for cognitively healthy participants, patients with mild cognitive disorder and patients with AD will be characterized to better our understanding of cognitive health’s link with cerebrovascular health. 
To summarize, the project is an exploration of the relation between the cerebrovascular system and the AD in order to comprehend the interactions involved and help with the diagnosis of the disease.
 

Interpretability of Artificial Intelligence, that is the capacity of an expert to understand why a prediction is made, is of great importance in health analysis. Firstly, because it matters to understand why a decision is made by an algorithm when it has such impact on a person’s life. Moreover, in research, interpretable algorithms are useful because they often unveil new investigation path. 

This study aims to combine two supervised machine learning algorithms to optimize both interpretability and performance, for instance, with mathematical logic tools. This new algorithm intends to help better predictions by lightly increasing model complexity while preserving high interpretability. 

This algorithm is developed to analyze fat data, which are data with a lot of characteristics (features) but with few samples (observations). This type of data is recurrent in health data, mainly in genomics, metagenomics and metabolomics data, which are all state of the art in medical analysis. More precisely, we are interested in problems such as antibiotic resistance or long corona virus disease (COVID-19). 
 

Il est souvent difficile de partager des données dénominalisées entre différentes organisations et chercheurs en raison de contraintes éthiques liées à la confidentialité des répondants. Il peut ainsi s’écouler de longs mois, parfois même des années, entre la rédaction d’un projet de recherche et le début de l’analyse planifiée, ce qui limite la capacité des chercheurs à mener des travaux scientifiques de pointe au moment opportun et contribue à allonger inutilement la formation d’étudiants gradués, entre autres problèmes. Une solution possible est de créer un jeu de données synthétiques à partager aux chercheurs en attente de l’accès au jeu de données original. Ce jeu de données synthétique serait représentatif des données originales, mais créé de façon à ne pas révéler d’information confidentielle sur les répondants. Il permettrait aux chercheurs de se familiariser à l’avance avec les variables mesurées, d’anticiper les difficultés techniques du projet de recherche (stockage, logiciels, gestion des accès), et de planifier de meilleurs protocoles de recherche.

Nous étudions ici les enjeux techniques liés à la création de tels jeux de données synthétiques dans le domaine de la santé. Il faut notamment s’assurer que les modèles statistiques utilisés soient assez flexibles pour bien modéliser les corrélations entre les variables collectées, tout en s’assurant de ne pas sur-ajuster ceux-ci, ce qui pourrait nuire à la protection de la confidentialité. Le travail s’articulera autour de la création d’un jeu synthétique pour un sous-ensemble des données collectées par le Consortium d’identification précoce de la maladie d’Alzheimer - Québec (CIMA-Q), pour qui le partage des données à la communauté de recherche sur la maladie d’Alzheimer canadienne et internationale est un objectif important.
 

The first objective of the project is to design efficient convolutional network classification models (CNNs) using mass spectrometry data (1D and 2D) for clinical diagnosis (cancer and infection).

Once finalized, the second objective is the interpretation of these classification models in order to identify spectral regions of interest that may correspond to new diagnosis or therapeutic biomarkers.

Radiotherapy treatments currently used in the clinical field are rarely modified. They generally consist of a global therapy of 50 grays, fractionated in five treatments of two grays every week for five weeks.
Thus, it could be worthwhile to develop a numeric tool, based on mathematical models found in the literature, in order to compare different types of treatment without having to test them on real tissues. Several parameters are known to alter the tissue response after irradiation including oxygen
partial pressure in irradiated regions, particle type hitting the tissue, and treatment duration.

The Python code created as the main part of the project is intended to facilitate the optimization of radiotherapy treatment by generating graphs showing cell survival after a certain number of fractions, taking many parameters into account. When completed and integrated to a graphical interface, the code will be easy to use and helpful for ongoing research projects.

Metabolomics is one way of studying metabolism. The presence of certain metabolites, or the breakdown of metabolic pathways can serve as indicators of a patient's health. They can serve as markers for certain diseases such as cancers, or provide information on the quality of an individual's diet. Untargeted metabolomics acquisition methods produce large data matrices. The aim is to develop machine learning methods specifically suited to handle high dimensional data sets. For example models based on decision rules.
The purpose of these models being the search for biomarkers, they must be sparse in order to be able to be interpreted by a human expert. We also try to develop new approaches to better interpret some existing and efficient models. Interpretability is essential in the application of machine learning to health. Models cannot be diagnostic black boxes but rather analytical tools available to experts to better understand human metabolism. 

It is often difficult to share denominalized data between different organizations and researchers due to ethical constraints related to respondent confidentiality. This is a common reality in the healthcare field, given the inherent sensitivity of this type of data. One option in this case is not to share the data directly, but rather to provide access to it via a tool that controls the risk of disclosure of the queries made and allows only those it considers safe. DataSHIELD is such a tool which has been proposed to protect the confidentiality of a dataset, and which can be used via the statistical software R. It also allows statistical analysis on several datasets hosted in different locations, always ensuring the confidentiality of the respondents. 

In this project, we are interested in the confidentiality guarantees provided by the software, and in its limitations. In particular, we wish to establish principles to guide the choice of disclosure control parameters offered with the tool, and to understand more precisely the impact of these controls on the quality of the descriptive statistics, linear models and graphs produced.
 

Data sharing is often limited by privacy issues. This is very common in particular for health datasets, given the inherent sensitivity of this type of data. When sharing of the original dataset is not possible, one method that can be used is to generate a synthetic dataset, which contains as much statistical information as possible from the original dataset, but which provides data on false individuals in order to protect the confidentiality of respondents. One way to ensure that these synthetic data effectively protect respondents is to use differential confidentiality, a rigorous measure of disclosure risk. 
This project is interested in how to analyze these synthetic datasets to obtain valid statistical results, as traditional methods of inference need to be modified to account for the variability added by the generation of the synthetic dataset.
 

This research project is based on the analysis of massive data on the NOL index and other intraoperative clinical parameters used by anesthesiologists during surgery. These parameters help them make analgesic treatment decisions in a non-communicating patient under general anesthesia and in whom it is impossible to assess pain and analgesic needs by standard questionnaires performed on awake patients. 
First, the objective is to interpret the values of this index in relation to the decisions made by the clinician. 
The second step is to develop an artificial intelligence algorithm that can guide decision-making for greater precision and better anesthetic safety for the patient.
 

Data sharing is often limited by privacy issues. This is very common in particular for health datasets, given the inherent sensitivity of this type of data. When sharing of the original dataset is not possible, one method that can be used is to generate a synthetic dataset, which contains as much statistical information as possible from the original dataset, but which provides data on false individuals in order to protect the confidentiality of respondents.

This project is interested in rigorously measuring the confidentiality protection offered by a synthetic dataset. We will carefully examine some measures proposed in the literature, to understand their guarantees and the differences and similarities between them in order to identify the measure (s) that would be the most relevant for the sharing of synthetic data.

Multipoint scintillation detectors are used to measure the dose of radiation deposited simultaneously at many locations in space and they have the advantage to allow real-time measurements. However, this detector must be precisely calibrated to provide accurate dose measurements.

The goal of this project is to develop an automated routine for the calibration of multipoint scintillation detectors under the beam of a linear accelerator such as the ones used for cancer treatments, by representing the calibration data in the principal component space.

A multipoint scintillation detector measures the spectrum of the light produced within the detector. Indeed, light is produced within the detector proportional to the radiation deposited in the detector. From a calibration dataset, a Non-Negative Matrix Factorisation algorithm (NMF) is used with the aim to retrieve the pure spectral components of the measurements. To simplify the visualization of the calibration dataset, the dataset is transformed using the Principal Component Analysis algorithm (PCA), and this transformed dataset is then represented graphically in the principal component space. This space allows to visualize the spectral composition of the data, relative to the pure spectra.

Many datasets can therefore be built, represented into this space, and used with the NMF algorithm with the aim to evaluate the performance of this algorithm for different calibration datasets.

In the end, this will allow to determine the experimental datasets that have to be acquired to perform an accurate calibration of the multipoint scintillation detectors.

This project studies the consequences of artificial intelligence (AI) systems and data science on public discourse, as well as their usage by the new content providers on the Web. 
It will tackle the ethical aspects of learning algorithms and recommendation filters implemented by internet companies to select and present content to the user. Specifically, the project investigates the consequences of such algorithms on public health, especially in the propagation of medical misinformation and pseudo-medicine.

This project aims at taking a critical oversight on data science techniques and their use. Various knowledge from different fields of humanities and social science will be applied (ethics, communication studies, philosophy of techniques) and will guide the development of technical solutions, as well as recommendations for the implementation of ethical and sustainable AI. 
For this reason, we will need both technical and philosophical research, working towards interdisciplinary integration.
 

The project focuses on the design, operationalization and validation of a sustainable health evaluation model.
This model will be adapted to a digital platform and based on solid theoretical and conceptual foundations. Furthermore, it will gather valid indicators and will be supplied by data showing a global and ecosystem conception of health.
Once operationalized, implemented and validated in a cohort study, this model will represent an innovative strategy for sustainable health through improved technologies and intervention methods.
 

This project aims to create data pipelines in diagnostic radiology in order to supply analysis and visualization tools.

The first pipeline is intended for data anonymization according to standard DICOM while the second one allows to supply Kibana (Elasticsearch) or Superset (Apache) platforms.

The Airflow orchestrator (Apache) is used to automate the execution of pipelines which could eventually supply dynamic dashboards. 
 

The research project is about the suitability of laws, legal principles and general framework surrounding health-related data, including those regulating the involved medical liability, in Canada and in the European Union. 
It aims to identify its weaknesses and aspires to provide regulatory solutions that are more appropriate to the realities of artificial intelligence. These solutions should better balance private and public, individual, social, commercial and health-related interests at stake. Also, this project considers a different view of the law and of our current legal systems with missing satisfactory answers.
 

This project focuses on the use of machine learning techniques in external radiotherapy for cancer treatment planning.
Stochastic frontier analysis is a parametric approach used in econometrics and appropriated for medical physics. Using a retrospective bank of treated patients it will be possible to predict the optimal dose of radiation for tumor and healthy organs.
This method is applied to multiple cancer treatment sites which emerge new challenge in the context of prediction, and data processing.

The characterization of Drug-Drug interactions (DDIs) is crucial for planning therapies and drugs
co-administration. While considerable efforts are spent in labor-intensive in vivo experiments and time-consuming clinical trials, understanding the pharmacological implications and adverse side-effects for some drug combinations is challenging. The joint impact of the majority of combinations remains undetected until therapies are prescribed to patients. This raises the need for computational tools predicting DDIs in order to reduce experimental costs and exhaustively characterize all drug combination effects before therapy recommendations. 
Previous attempts to build such tools focused on pharmacodynamic and pharmacokinetic interactions and used features that are difficult to access in the early stages of R&D. 
In this work, we propose to use data about the drugs and their targets (pathways, biomarkers, gene expressions, etc) that are available at the beginning of each drug R&D campaign. Our hypothesis is that high-level deep learning features extracted from those data will improve DDI characterization. Therefore, our models will be trained to output the pharmacological effects of DDIs as well as underlying molecular and biological pathway interactions. 
Creating such a comprehensive toolkit will help to reduce risks in polypharmacy therapies.
 

The research project is focused on the synthesis of medical images using deep learning, towards better artifact correction and the avoidance of unnecessary medical procedures.

The neural networks designed in this project have a flexible architecture enabling the image synthesis from only an heterogeneous subset of input modalities. The images are synthesized in pathological situations, such as Alzheimer's disease and brain cancers.

This project is centered on an examination of the process of preparing a patient-centered media and social media strategy that provides patients with useful information about the Opal patient portal and how they can make the most of it.

This project is part of the effort to prepare Opal for use by patients, providing them with self-management resources such as questionnaires and educational materials.

The primary objective of this research project is to detect cancer pain at an early stage by analyzing patients’ medical images. 
Development of an algorithm to do this can be achieved by combining two computer science techniques: one that allows us to gather information about pain from medical notes, and one that extracts information from medical images. We will use the first technique in a computer program that will extract and quantify pain intensity recorded in patients' medical notes. 
The second technique will be employed in another program that will analyze radiographic images of cancer patients’ to extract information about their bone metastases (such as tumor volume, and shape). Then, we will implement advanced statistical and mathematical techniques to model the relationship between identified tumor features and extracted pain intensities. 
Finally, to validate our model, we will use pain scores that are directly collected from thousands of future cancer patients via a mobile app that has been developed in our group (opalmedapps.com).
 

This research project aims to examine the mutational signature of ionizing radiation using single-cell sequencing techniques.

The project is using human lymphoblastoid cells donated by the Ashkenazi trio that have a well characterized genome. The cells are irradiated and sequenced to determine the mutations that are induced as a result of the exposure to the ionizing radiation.

Through biostatistical analysis of the human genomic data thus obtained, we will be able to identify the mutational signature of ionizing radiation.
 

A patient portal is an extension of an electronic medical record system that is accessible to patients. Although patient portals have been around for many years, they have had poor adoption in Canada. This is due in large part to the desire of provinces to invest in large centralized electronic medical record systems and the complexity of implementing such systems. But patients are demanding access to their medical data and do not wish to wait for complex centralized systems to be implemented.

Therefore, in this research project, we will expand and evaluate the Opal patient portal, previously developed and implemented at the McGill University Health Centre, to function as a multi-institutional patient portal using a novel asynchronous data federation infrastructure.

This project is centered on the data donation aspect of Opal and will involve a Privacy Impact Assessment of the application and the eventual practice of data sharing driven by it.

This project involves two components: (1) preparation of Opal for the caregiver functionality in which patients will be able to share some or all of their medical data with their caregivers, and (2) general content preparation for Opal.

This project involves an examination of the regulatory privacy and confidentiality compliance requirements for the use of a patient portal in various Canadian provinces.

Romina also works as a member of the quality assurance team, the market research team, and assisted with deploying Opal in numerous clinics within the Cedars Cancer Centre.

This research project is focused on preparing a pilot project for the donation of radiotherapy data by radiotherapy patients using the Opal patient portal.

This project is investigating ways in which patients can share their data and it will put in place the infrastructure for a demonstrative project.

This research project is focused on using blockchain or an alternative solution to provide security for data donation using the Opal app.

It will put in place a demonstrative blockchain infrastructure, examining its challenges and drawbacks and proposing potential innovative solutions.

This project is focused on determining if patient-reported outcomes are better correlated with actual dose delivered than with planned dose in prostate cancer patients receiving radiotherapy.

The project will use daily cone-beam CT images to calculate the daily and total radiation dose delivered to patients, and the Opal app to collect their patient-reported outcomes. 

Process efficacy and robustness are crucial to assure productivity and predictability in pharmaceutical manufacturing. Medicago’s vaccine manufacturing technology uses plants for production and our aim is to develop a system capable of predicting and monitoring plant’s fitness for production early in the process, from plant seedling to harvest of producing leaves.

To this end, we must identify the factors that regulate the production level for each plant. We plan to measure a great deal of molecules, called metabolites, to determine the optimal conditions for the plants to generate the maximal amount of each product. Since the quantity of measurements is large, we will use machine learning to design an artificial intelligence capable of understanding and identifying the potentially highly complex patterns of metabolites and/or biomarkers that are correlated with the optimal production.

The discovery of antimicrobial agents was one of the great triumphs of the 20th century. The sobering news is that antibiotic resistance was part of the process as well. If nothing is done by 2050, antimicrobial resistance  (AMR) will cost $100 trillion with 10M people/year expected to die (https://amr-review.org). Factors driving AMR extend beyond human healthcare with implications in veterinary medicine, agriculture and the environment (the One Health approach). New and improved approaches for tackling AMR include better surveillance; rational drug use, different business model for generating antibiotics, innovation at all levels and most importantly a global approach.

This transnational team grant proposal is tasked to apply new machine learning approaches for modelling AMR for faster diagnosis, better surveillance and prediction of resistance emergence. Specifically, we will develop machine learning implementation that can orient the selection of treatments by assessing the level of resistance, provide rational for the generation of novel antibiotics, and assist in the surveillance of human and livestock AMR around the globe.

To achieve this, we have assembled a transnational team (Canada, China, Finland, France) with complementary skills with demonstrated expertise in machine learning applied to both genomics and metabolomics and AMR domain experts. Our transnational team has all the elements to be highly impactful and to continue collaborating well past the JPIAMR funding period.

The complementarity of our expertise will help us to tackle the challenges ahead and ensure our continued success. 

The aim of this doctoral thesis is to develop a tool able to automatically provide organs of interest segmentation in computed tomography images using machine learning techniques.

This tool will then be used to calculate organ doses in order to establish personalized dosimetric records in medical imaging. Doses will be calculated using informations obtained from images, radiographic technique and GPU-based Monte Carlo dose calculation algorithm (GPUMCD). Automated pipelines will be implemented to process large amounts of data.

The proposed tool provides a better evaluation of population exposure to ionizing radiation caused by medical imaging procedures.

Prostate cancer is the most common form of cancer in men in Canada.

This research project aims to establish a prognosis for a patient suffering from prostate cancer as well as predict the final pathology, by predicting the presence of lymph node metastases, from a FDG PET-CT. Radiomic characteristics are defined as the process of quantitative extraction of usable high-dimensional data from medical images. These are biomarkers that are difficult to see with the naked eye, such as texture and intensity. The database is made up of 250 prostate cancer patients. After filtration, a subset of 331 radiomic characteristics was selected. The accuracy of the model is 74.5%. This corresponds to an increase in precision of 6% compared to a model trained on all the extracted characteristics.

Ultimately, the algorithm will better predict the risk of recurrent prostate cancer and help improve methods and choice of treatment.

The clinical and economic burden of prostate cancer in Canada is substantial and is rising. It has been indicated that 1 in 7 men will develop prostate cancer during their lifetime, and another 1 in 27 will die due to the prostate cancer. However, only a part of prostate cancer cases is clinically important which make the prostate cancer case discrimination crucial to avoid over-treatment. Compared to ultrasound imaging method, advanced MRI modalities have demonstrated a better diagnostic accuracy and is becoming a clinical routine examination for patients at risk of having clinically significant prostate cancer. Even though the version two of PI-RADS has been recently published to facilitate MRI modalities application in prostate cancer, they still present limitations. For instance, variability is reported in terms of inter-reader agreement and diagnosis accuracy, mainly depend on reader experience.

This project aims to find a machine learning based approach for predication and segmentation of intraprostatic lesions to better guide radiation treatment. For accomplishing this task, the most advanced MRI modalities including DTI-MRI and DWI-MRI along with the anatomical MRI modalities will be employed. From the quantitative MRI modalities several maps that enhance specific features of the lesion will be extracted. Then after, texture information of the MRI modalities and selected maps will be extracted. In this step machine learning methods will be employed for feature selection and classification purposes. Finally, the prostate cancer extension and its type are identified.

This project consists of establishing good practices in health data management and building a software infrastructure in order to apply them.

We have developed pipelines that allow daily recovery of brachytherapy treatment data in order to calculate and store their dosimetric indices in a database dedicated to research. These indices are essential for planning radiotherapy treatments and for estimating their quality.

The aggregation of these indices allows different researchers such as bio-statisticians and radiation oncologists to carry out studies on larger data sets.

The objective of this project is to extract a set of relevant data from the files produced by medical imaging devices.

The process consists of building ETL (extract-transform-load) pipelines to make the data consumable for analysis and visualization. An example of analysis consists in observing the trend in dose administered to patients according to the establishment, protocol or device used, in order to possibly identify non-standard practices.

The data extracted could also guide practice by making it possible to assess the relevance of certain studies, and thus to optimize resources in the health network.

An exception drug is a drug that is not usually covered by the public drug insurance plan (RPAM). The measures implemented at the RAMQ for exceptional drugs allow the entire population to obtain coverage for certain drugs if they are used in compliance with the indications recognized by the Institut national d'excellence en santé et services sociaux (INESSS). Exception drugs are now a large and constantly increasing part of total spending on prescription drugs.

For the RPAM, one of the ways to control this increase is to reimburse these drugs according to pre-established rules. Currently, the system automatically processes around 20% of requests while the rest are directed to a case-by-case analysis, which generates delays.

This project is to help the business sector respond more quickly to requests for approval of exception drugs. A tool will be developed based on 15 years of data collected by the current system, and will aim to increase the number of requests processed automatically.