The rapid development and deployment of artificial intelligence demands we connect our best and brightest minds, and work to train the next generation of research leaders. In June, the Fonds de Recherche du Québec Santé (FRQS) announced $4.5 million for three , all three of which were awarded to teams co-directed by McGill researchers. The program brings together researchers with complementary expertise in AI, data sciences and life sciences to address issues and challenges impacting the health of Canadians and the efficiency and effectiveness of the healthcare system. With the investment from this and a previous call in 2021, the program will facilitate simultaneous research training for more than 60 students and postdoctoral fellows in the fields of AI and life sciences.
Each chair will receive $1.5 million, distributed over three years. The Dual AI Chairs are supported in part by the ministère de l’Économie, de l'Innovation et de l'Énergie. As of July 1st, the programs are actively recruiting trainees.
“With this significant support from Le Fonds de recherche du Québec – Santé (FRQS) an emerging generation of researchers will develop the skills and expertise they need to design the health solutions of the future, to make medicine safer, and to advance treatment for some of the most devasting diseases and disorders,” said Martha Crago, Vice-Principal, Research and Innovation. “The fact that McGill researchers are co-directing all three FRQS Dual AI Chairs is truly impressive, and a testament to the expansive expertise and collaborative spirit of our AI, data sciences, and life sciences research communities,” she added.
Professor of Neurology and Neurosurgery and Director of the Centre for Research in Neuroscience (RI-MUHC), Keith Murai, and McGill Professor of Computer Science, Kaleem Siddiqi, will co-direct the Dual AI Chair, Cracking the nanoscopic structural code of the brain: Artificial intelligence and computer vision approaches for brain health, which promises to advance understanding of Alzheimer’s and other neurodegenerative diseases.
McGill Associate Professor of Medical Physics, , and Associate Professor in the at Polytechnique Montréal, , will co-direct the Dual AI Chair, Smart data for smart cancer care – a research program that combines expertise in natural language processing, semantic web technologies, and patient-centered data to create knowledge bases in oncology. With the goals of reducing risk and making cancer treatment safer and more effective, Kildea and Zouaq are collaborating to build an AI solution that will combine, consolidate, and exploit unstructured health data.
Mathieu Blanchette, Associate Professor in McGill’s School of Computer Science will co-direct the Dual AI Chair, DĂ©veloppement d’approches en intelligence artificielle pour Ă©lucider les codes de rĂ©gulation des ARN et exploiter leur potentiel thĂ©rapeutique, with of the MontrĂ©al Clinical Research Institute (IRCM) with This program aims to tap into the potential of AI to facilitate discoveries in Ď㽶ĘÓƵ biology and therapeutics.
Cracking the brain code:
The ability of the human brain to process, encode, and transmit information relies on its complex cellular interactions. When communication across cells is interrupted, it can lead to debilitating conditions such as Alzheimer’s disease. Yet developing effective treatments for Alzheimer’s and other neurodegenerative conditions is limited, in part, by our incomplete understanding of the complex three-dimensional structure of brain cells at the nanometer scale.
To address this knowledge gap, Professors Murai and Siddiqi, have joined forces to develop first-of-their-kind, artificial intelligence (AI)-driven tools to analyze the nanostructure of the brain. Their dual chair program will build AI and computer vision tools to elucidate the organization and interplay between brain cells and their cellular components. “We're focusing the research using a strong fundamental perspective in terms of answering some very simple questions about cellular shape and complexity and its implications for brain connectivity patterns,” said Murai.
These tools will depart in fundamental ways from the AI models currently in the public domain, which are built for classification tasks or detection tests. To model the three-dimensional structure of brain cells at the nanometer scale, the AI models must be more complex, argued Siddiqi. “It's not just about input-output, but about understanding relationships, and ultimately, it has to be connected to biological processes,” he said.
The innovative models they are pursuing hold great promise: clinicians will have a greater ability to detect early disease-driven structural changes to brain architecture; researchers will have the information they need to develop new interventions to stop or slow the progression of devasting neurodegenerative conditions; and, further down the innovation pipeline, therapeutics can be developed to improve the life of individuals suffering from devastating diseases.
The student training component of the program is top of mind for Murai and Siddiqi, who view it as an amazing opportunity to help an emerging generation of researchers gain knowledge in more than one field. It is not lost on them that their research partnership—and shared lab space—is due, in part, to the keenness of a student who was enrolled in their classes. Seeing cross over potential in their work, the student insisted that Murai and Siddiqi meet and talk about “some cool problems” they might jointly tackle. Siddiqi therefore hopes that for the up-and-coming researchers in their labs, “it will be natural to learn computer science and neuroscience simultaneously.”
Better cancer care with AI:
In almost all healthcare settings in Canada, a digital chart accompanies a cancer patient’s appointment with a physician. Each specialist who sees a patient reads existing information about the patient before their encounter and writes a digital note about the visit after. The note contributes some new information to the patient’s chart but, more-often-than-not, it also repeats—sometimes erroneously—redundant information available elsewhere. The result is that for many cancer patients, important information is buried in clinical notes in isolated parts of their electronic health record. Useful information, that could otherwise contribute to advances in cancer care and cancer research, remains inaccessible. An AI solution to combine, consolidate and exploit unstructured health data is needed.
This is the wicked problem that the Dual AI Chair led by Kildea and Zouaq aims to solve. “There is so much data now that clinicians and researchers have to manage,” says Kildea. “Our project aims to help the clinician and the researcher find the pertinent information that they are looking for.” He added that, “in this way, AI can reduce risk and ultimately make cancer treatments safer.”
The Kildea lab has a track record of using agile methods to build award-winning patient-centered digital health applications. In collaboration with a patient (the late , who was also a professor of Computer Science at McGill) and RI-MUHC clinician (Dr. the lab created the . Opal was named Quebec eHealth solution of 2019 and awarded the 2019 Prix d’excellence–Coup de coeur des ministers, the highest accolade of Quebec’s Ministry of Health and Social Services. Kildea also created the (QSCC), which is advancing the use of patient-centred data and mobile health technologies for remote monitoring and AI research in healthcare.
Zouaq’s LAMA-WeST (Web, Semantics and Text) lab works at the cutting edge of natural language processing (a branch of AI that enables computers to comprehend, generate, and manipulate human language) and Semantic Web technologies, which enhance knowledge exchange, information management, and decision support in healthcare systems. By working together, they will bridge a research divide: for Kildea, access to AI expertise, and for Zouaq, access to real world data from healthcare settings.
Driving progress in AI healthcare requires this close collaboration and the results of their research partnership have the potential to provide new approaches to cancer research and care, support clearer communication among clinicians, and link massive volumes of available healthcare data. For Kildea, this work is inherently collaborative. “We're building on work that has gone on for many years,” he said. “Now, it's about connecting with our colleagues that can complement our work.”
A Made-in-Quebec Ď㽶ĘÓƵ-based therapeutic revolution:
Ď㽶ĘÓƵ defects frequently cause diseases, including cancer and incurable spinal muscular atrophy and Huntington’s. Understanding the mechanisms causing Ď㽶ĘÓƵ regulatory defects is therefore critical to developing new therapeutic approaches and treatments. The field is burgeoning, and scientists have achieved some major successes using Ď㽶ĘÓƵ as therapeutic effectors or targets for previously untreatable and undruggable diseases; and yet, new advances are needed to speed up the development of new treatments. For example, while mĎ㽶ĘÓƵ vaccines have been gamechangers in the context of the Covid-19 pandemic, they have yet to be viable for treating diseases such as cancer.
“There's a dynamic flow of ideas and of data that is making AI approaches very powerful for questions relating to Ď㽶ĘÓƵ biology” said Blanchette. “This Dual AI Chair enables us to target some of the most promising biomedical applications, with a direct impact on both fundamental research and biotechnology and drug development,” he added.
The potential of AI to decipher Ď㽶ĘÓƵ regulatory codes and to extract interpretable information on Ď㽶ĘÓƵ sequence, structure, and post-transcriptional regulation, is as great as the tasks are daunting. The Dual AI Chair led by Blanchette and LĂ©cuyer will therefore tackle these challenges from multiple angles and across three themes: developing machine learning AI approaches to elucidate Ď㽶ĘÓƵ regulatory mechanisms; designing AI-guided strategies to understand disease-associated Ď㽶ĘÓƵ dysfunctions; and using AI to benefit Ď㽶ĘÓƵ-based therapeutic strategies.
Quebec is proving an ideal research and training ground for Blanchette and LĂ©cuyer. This is a province where Ď㽶ĘÓƵ biochemists have long worked together with computer scientists and mathematicians, and there is an established a network of experts in AI and biochemistry. This nexus of cross disciplinary expertise forms the backbone of many large-scale research initiatives, including McGill-based D2R, which recently received $165 million through the Canada Research Excellence Fund (CFREF) to pursue genomically driven Ď㽶ĘÓƵ-based therapeutics. Quebec’s AI revolution is equally impressive, evidenced in part by the joint creation by UdeM and McGill of Mila (the Quebec AI Institute) in 2019, and the launch of UdeM-based CFREF-funded IVADO.
Among the predicted developments and outcomes of this Dual AI Chair are novel approaches to predict higher level phenotypes such as Ď㽶ĘÓƵ stability, localization and translation in human cells and new predictive models to synthesize existing research, as well as new predictive neural network models, and better strategies to design mĎ㽶ĘÓƵ vaccines, including for cancer. Blanchette echoed that one of the most important outcomes will be a newly trained cohort of students working at the interface of two of the fastest-growing areas in science and biomedicine. “We want to solve real world problems, and these partnerships ensure that we address the most important questions,” he said.