Data centres – large industrial facilities that power cloud computing and AI – have become critical infrastructure supporting the world’s growing digitization. Everything from streaming video and online banking to scientific research and generative AI depends on their ability to store, process and move enormous volumes of data.

As demand for digital services continues to rise, these centres sit at the root of that growth. And, as they become more pervasive, conversations about broader implications are growing.

“Data centres are increasingly part of public debate because of concerns about energy use, environmental impact, local economic effects and data sovereignty in Canada,” says Lyndsey Rolheiser, an assistant professor at the .

Despite the growing significance, there remains a notable gap in publicly available information about these facilities.

“There is very little systematic evidence to inform that discussion,” says Alexander Carlo, a postdoctoral researcher at Schulich. “At a basic level, we do not have a clear picture of where data centres are located in Canada or where new ones are being developed.”

Rolheiser and Carlo set out to address that gap by creating the first comprehensive map of Canada’s data centre landscape. Their work, now and to be included in the forthcoming Schulich School of Business Real Assets Research Paper Series, documents both existing facilities and the growing pipeline of projects that have been announced or are under construction.

The authors built their analysis around a proprietary dataset from Aterio, a data intelligence firm that aggregates information on large‑scale infrastructure projects. Using permitting records, utility filings and company disclosures, they tracked facilities from initial announcement through construction to full operation, then layered in census and provincial electricity data to assess location, scale and energy implications.

Once completed, they mapped out a much clearer picture of how Canada’s digital infrastructure is changing. The analysis shows that while Canada’s current data facilities footprint remains relatively modest, the pipeline of planned facilities is nearly 10 times larger – and those new centres are far bigger than older ones, reflecting a shift toward hyperscale infrastructure designed to support AI.

Future development is also highly concentrated: Alberta alone accounts for more than 90 per cent of planned capacity, despite relying on a comparatively high‑emissions electricity grid. At the same time, new facilities are increasingly being built far from major cities, often hundreds of kilometres from urban cores. Meanwhile, provinces with cleaner electricity systems, including Quebec, Ontario and B.C., have begun restricting or carefully managing grid access for large new data centres.

These patterns reflect a set of broader concerns the authors explore in the paper. Data centres consume enormous amounts of electricity – often equivalent to tens of thousands of households per facility – while creating relatively few long‑term jobs compared with the scale of public infrastructure they require. Their expansion can reshape provincial power systems, raise emissions concerns and crowd out other users. The authors also point to questions of data sovereignty, since most large facilities are owned by foreign firms and to the risk that some projects could become stranded assets if AI demand slows or climate policy tightens.

While Rolheiser and Carlo do point to these risks, the aim of the research is to ground future discussions in evidence. “This is a necessary first step for any informed policy or public debate,” Rolheiser says.

“At a minimum,” Carlo adds, “the paper should help clarify what the current landscape looks like and where development is taking place.”

Both researchers hope their work contributes to more informed discussions about data centres in Canada, and provides a solid evidence base that helps policymakers and the public better understand these sites and their impacts on grid access, emissions and economic benefits.

The post Researchers at York create first map of Canada's data centres appeared first on YFile.

The study, published in , tested a multi-agent system built with large language models (LLMs) to simulate the work of a liver transplant selection committee – a multidisciplinary group that decides which patients are placed on transplant waitlists.

Using real-world transplant registry data, the AI system demonstrated high accuracy in identifying patients who are likely to benefit from a liver transplant and those for whom transplantation would be unlikely to help.

“Liver transplantation is a rare case in medicine where access to a life-saving treatment is limited by organ availability,” explains co-senior author Divya Sharma, assistant professor in the Faculty of Science. “Decisions about who is waitlisted are complex, and committee deliberations can be subject to unconscious bias where a clinician's own background or identity may subtly influence their judgement, even when national guidelines are in place.”

Researchers set out to test whether AI agents – each assigned a clinical role – could support more objective decision-making. To test the approach at scale, researchers evaluated the system against transplant outcomes data.

The study analyzed 20 years of data from more than 8,000 adult liver transplant recipients in the U.S. using the Scientific Registry of Transplant Recipients. A simulated group of patients with known contraindications was also created to test the system’s accuracy in flagging cases that should be excluded from transplant consideration.

Results show the AI committee predicted one-year post-transplant survival with 92 per cent accuracy and six-month survival with 95 per cent accuracy. Contraindications were identified with an accuracy of more than 98 per cent, thereby identifying transplant candidates efficiently.

The research team also examined where errors occurred to better understand where the AI system works well, and where it needs careful oversight and improvement. The authors caution that continued monitoring is needed because transplant data can reflect broader inequities in access to health care.

“Our work positions LLM-based multi-agent AI systems as potential clinical decision-support tools, rather than replacements for human judgement,” says Sharma. “While AI shows promise in making liver transplant decisions more objective, it’s crucial to emphasize that the final responsibility must always remain with transplant teams and human oversight is critical to address ethical considerations.”

Sharma says while more research is needed to test the AI tools in real-world settings across different health systems, AI-supported committees have potential to help standardize complex medical decisions where resources are limited.

The post Can AI reduce bias in liver transplant waitlists? appeared first on YFile.

Several times a month, a small drone rises above the trees at Swan Lake, following a precise path over the water. Parkgoers who enjoy walking, jogging or birdwatching might assume it’s there to capture scenic footage. Instead, the drone is part of a �첥��Ƶ-led effort to understand – and help restore – the health of an urban lake under pressure.

Swan Lake, a former gravel pit transformed into a stormwater pond and community green space, faces ongoing water quality challenges. As rainwater flows into the site from surrounding roads and neighbourhoods, it carries excess nutrients, road salt and other pollutants. Over time, this can fuel frequent algae growth, cloud the water and reduce oxygen levels, stressing fish and wildlife, limiting recreation and, in some cases, raising public health concerns.

Since April 2025, �첥��Ƶ researchers, led by CIFAL York, have been turning concern about the lake’s health into measurable data and practical action through the Swan Lake Citizen Science Lab (SLCS Lab). The initiative brings together York research centres, including ADERSIM and the One WATER Institute, with local partners such as Friends of Swan Lake Park, a community‑based volunteer organization dedicated to protecting and improving the area’s ecological health.

“Communities often know when something is not right with a local ecosystem, but it’s hard to act without clear, comprehensive and consistent information, as well as meaningful community engagement” says Ali Asgary, director of CIFAL York and professor in the Faculty of Liberal Arts & Professional Studies. “The goal of the lab is to support those concerns with reliable data that can guide real decisions.”

"To assess a lake is to assess ourselves," adds Satinder Kaur Brar, director of the One WATER Institute and professor at the . "Its health card is a mirror of our environmental stewardship."

One way the lab is assessing the lake is through advanced technology, such as the use of multispectral and thermal drones operated by York research teams.

Equipped with special cameras that capture different types of light – including some invisible to the human eye – the drones can detect potential algae growth and subtle changes in water clarity as they scan the lake from above. Flying low and on demand, they provide detailed, up-to-date views of trends across the entire water body, offering a clearer picture than satellite images and a broader perspective than scattered and spot‑by‑spot water sampling.

The drones have already yielded valuable insights, recently shared in a York‑led, under-review study that monitored patterns from spring through fall 2025. By flying the drones roughly once a month and analyzing the findings over time, researchers were able to pinpoint where algae forms, how blooms shift across the seasons and how changes in water cloudiness are driven by biological growth rather than stirred‑up sediment.

The findings confirm what many residents and park managers have long suspected: the lake is rich in nutrients and prone to recurring algae growth. The drone data, however, also reveal something new.

Conditions vary significantly from one area to another, suggesting that targeted, location‑specific interventions may be more effective than broad, one‑size‑fits‑all treatments applied across the entire lake. Knowing where problems emerge helps guide chemical treatments, shoreline naturalization projects and future restoration efforts – and provides a better way to measure whether those interventions are working. "Interconnecting drone data with on-ground water quality can turn ecological signals into informed action that is vital for communities," says Brar.

“What the data made clear is that this isn’t a uniform problem,” adds Asgary. “When conditions vary so much from one part of the lake to another, it changes how you think about solutions. This kind of information allows us to be more precise, more proactive and more strategic in environmental management.”

In addition to monitoring Swan Lake, York‑led teams are working to make the data easier to interpret and use in planning. Researchers are developing AI tools to identify patterns in the drone imagery, anticipate conditions such as algae outbreaks and translate complex trends into clearer insights.

Other teams are using virtual reality and simulation to help users visualize the lake over time and explore how different interventions might affect conditions. Meanwhile, geographic information system (GIS) specialists are turning the results into interactive maps and dashboards that help the public and those involved in lake management understand what is happening across the site.

A core goal of the Swan Lake Citizen Science Lab is to encourage meaningful community engagement and shared stewardship.

“From the start, this was never about researchers working in isolation,” says Asgary. “The goal of the Swan Lake Citizen Science Lab is to create a shared process, where community knowledge and scientific tools come together.”

Local partners are not just observers; they are active partners in the research. Residents take part in field checks, help interpret findings, attend workshops and contribute to outreach efforts that share findings. Alongside them, �첥��Ƶ students gain hands‑on experience applying classroom learning to a real environmental challenge, working with researchers and resident members in a local setting.

For CIFAL York, which is affiliated with the United Nations Institute for Training and Research, the work at Swan Lake is a pilot that could inform other communities facing similar pressures on small urban lakes and wetlands.

“The impact here is very tangible,” says Asgary. “Through drones, data and collaboration, we’re building a deeper understanding of how this ecosystem functions and how it can be protected over time. That kind of shared knowledge is what allows stewardship to last.”

Find out more about the SLCS Lab, and see it in action, in the video below.

The post How York is helping to restore an urban lake appeared first on YFile.

Emergency management at airports is uniquely demanding because of the complex, diverse and dynamic systems involved, says Ali Asgary, professor of disaster and emergency management in the Faculty of Liberal Arts & Professional Studies.

With dense traffic, multiple vehicles and operations often unfolding during changing or extreme weather, coordinating airside and landside activity remains a major challenge.

“Even a small emergency at an airport can have significant political consequences and cascading impacts,” Asgary says. “These are the dynamics that shape airport emergencies, runway incidents and large‑scale disruptions to air transportation.”

Asgary's research has gained renewed relevance amid the March 22 Air Canada collision between an aircraft and a fire truck on a runway at LaGuardia Airport. While investigations are ongoing, the fatal incident underscores how seconds matter during runway operations.

While it’s still too early to determine what led to the tragedy, Asgary says events often involve factors that emergency managers and aviation operators routinely study: real-time hazard assessment, workloads, communication and warning systems.

“Runway incidents often involve overlapping risks, including split‑second decision‑making, heavy controller workload and limited redundancy in warning systems,” he says. “When warning systems rely on a single communication channel, missed messages can quickly escalate into serious incidents.”

Asgary is executive director of – the Advanced Disaster, Emergency and Rapid Response Simulation lab at �첥��Ƶ – where researchers and students simulate disasters and test response plans before they emerge in real‑world settings.

At ADERSIM, researchers use agent-based models to simulate aviation scenarios and examine how decisions by pilots, passengers, crew and ground emergency responders influence outcomes.

The lab incorporates virtual reality to help emergency managers visualize airport events and uses AI to analyze disruption patterns. It also explores how tools such as drones could support airside emergency response and risk assessment.

ADERSIM has also developed AeroHaz, a web-mapping application that identifies major hazards for airports worldwide to support hazard awareness and planning.

“Through a combination of computer modelling, human‑in‑the‑loop simulations, extended reality and AI, we can test how emergency response systems behave when multiple risks converge and conditions change rapidly,” says Asgary. “The work of ADERSIM contributes to York's leadership in disaster and emergency management.”

Major runway incidents can yield lessons for emergency preparedness – but only if they are researched, documented and incorporated into revised procedures. The incident also highlights the need for more research into the technological and human factors driving airport safety.

“Simulation-driven research allows emergency planners and responders to review how decisions are made, how workflows unfold in crisis situations and how to improve preparedness,” says Asgary.

In addition to leading ADERSIM, Asgary is also director of CIFAL York, a UNITAR centre that connects academia with leaders and organizations to tackle global challenges through specialized training in disaster management, sustainability, health and entrepreneurship.

Together with Maleknaz Nayebi, associate professor at the and associate director of CIFAL, he is leading a project to develop AI solutions for airports to minimize risks and enhance response operations. Using AI can help predict weather conditions, coordinate workforces and more.

ADERSIM and CIFAL York also share this research through training and professional learning for airport and emergency management leaders, and through public events.

Those who are interested in learning more can attend a two-part webinar series titled Airport Operations, Passenger Management, and Technology in the Face of Geopolitical Crises. Presented by CIFAL York and ADERSIM, in collaboration with UNITAR, the event runs April 15 and 25.

CIFAL York and ADERSIM will also contribute to UNITAR’s Airports Global Training Programme, when Nayebi will host “Future-Ready Airports: Preparedness for Mega Events Through Safety, Sustainability, and Smart Innovation” on April 22 and 23 in Atlanta, Georgia.

The post �첥��Ƶ simulation research supports airport emergency preparedness appeared first on YFile.

CIFAL York is expanding its work in climate innovation with a new focus on how AI can support real‑world solutions to some of the most pressing environmental challenges.

Since its establishment in 2020, CIFAL York, part of the United Nations Institute for Training and Research (UNITAR) global network, has been at the forefront of climate change, disaster management and sustainable development. It offers innovative approaches to climate challenges, including training on emergency management, workshops on disaster risk reduction and programs that help local leaders prepare for both climate and health crises.

With the rapid evolution of emerging technologies showing great potential to support efforts in climate solutions, the centre is now expanding its mandate. “We want CIFAL York to be a leader in exploring the intersection of AI and climate change,” says Ali Asgary, CIFAL director and professor of disaster and emergency management in the Faculty of Liberal Arts & Professional Studies.

Its first step toward that work is the launch of the Climate AI Innovation Hub, an initiative designed to explore how AI can support creative approaches to addressing climate challenges. Its goal, says Asgary, is to create a network for knowledge sharing, innovation and collaboration that can achieve real-world impact.

The hub’s first initiative – a monthly speaker series running until November – sprang from the idea of leading conversations that explore what is possible with AI.

“These computational powers can help us understand and analyze changes in climate. Maybe they can even prevent them by allowing for proactive – more than reactive – approaches,” says Maleknaz Nayebi, associate director of CIFAL and assistant professor in the . “It’s not that there is one answer that can be given. For us, it’s about raising those questions. That’s how we came up with the speaker series.”

The series will showcase, for example, how AI, IoT (the Internet of Things) and satellite technologies are being used to tackle pressing environmental risks – from predicting and managing wildfires to designing low-waste, circular buildings. It will introduce participants to the broader climate innovation ecosystem and highlight the role of innovators and entrepreneurs creating scalable solutions for sustainability, resilience, circular economies and low-carbon transitions.

The series will raise awareness about climate entrepreneurship, explore sector opportunities and obstacles, and empower students, early-career professionals, founders, researchers and community innovators to take an active role in environmental research leadership.

“Our goal is to help people understand how these technologies are being developed and used, and to encourage the sharing of innovations,” Asgary explains. “We hope to inspire the next generation of climate innovators and show potential users – particularly government agencies – what tools and solutions are available to them.”

The speaker events are the hub's first step in engaging the community, and Asgary says past CIFAL series have served as a foundation for building networks of researchers and practitioners through live group discussions. Recorded content available on also becomes a knowledge repository that draws in new audiences.

“Many of our research projects in recent years have been fed by our speaker series,” says Asgary. Other outcomes have included white papers, book chapters, courses, certificate programs, short courses, community events and more.

Feedback from the first session in February suggests the new series is cultivating projects informed by the insights and networks it generates, highlighting the promise of what CIFAL aims to achieve.

“The hub is about creating connections, sparking new ideas and ultimately applying AI responsibly to make a tangible difference,” says Asgary. “At the end of the day, the goal is to contribute to solving climate change.”

The post CIFAL York debuts hub to explore AI solutions for climate change appeared first on YFile.

SDG Month feature

As Earth’s orbits grow increasingly crowded with satellites and space debris, �첥��Ƶ researcher Zheng Hong (George) Zhu is developing technologies to keep space safe and sustainable.

In addition to the more than 11,500 operational satellites orbiting Earth, according to the Satellite Industry Association, tens of thousands of pieces of space junk – including retired satellites, discarded rocket components and metal fragments – now occupy Earth’s orbital environment.

Managing orbital debris has become central to the long‑term sustainability of space activity, as NASA prepares new crewed missions to the moon, cargo vehicles continue servicing the International Space Station and companies such as SpaceX’s Starlink deploy thousands of satellites.

“In the past, we thought of Earth’s orbit as having infinite space, but it doesn’t,” says Zhu, professor in the Department of Mechanical Engineering at York’s and Tier 1 York Research Chair in Space Technology. “If you don’t clean up, eventually it becomes very risky.”

As waste accumulates, the risk of collisions with satellites and spacecraft rises. This can eventually lead to the so-called Kessler Syndrome, where collisions trigger a chain reaction that produces more and more fragments which, in turn, cause additional collisions.

Even small objects travel at extreme speeds: a fragment no larger than a bolt can disable a satellite that supports services people rely on every day, from global communications and weather monitoring to navigation and emergency response.

Without effective strategies to reduce and remove debris, humanity’s ability to safely operate in space may be compromised.

Zhu has spent more than a decade developing solutions to that challenge.

His interest in space debris mitigation goes back to 2010, when he says few researchers were focused on the issue. Two high‑profile events sharpened his attention: China’s 2007 antisatellite missile test – which destroyed one of its own aging weather satellites and scattered thousands of fragments – and the 2009 accidental collision between an operational U.S. communications satellite and a defunct Russian satellite, the first known crash between two intact satellites in orbit.

“It was a wake‑up call that caught my attention,” says Zhu.

He began exploring how a technology he was studying, called electrodynamic tethers – long, thin conductive wires that interact with Earth’s magnetic field – could help address the problem. Originally investigated as a way to generate electricity in orbit, Zhu realized the technology could also act as a brake, slowing satellites and objects so they safely re‑enter Earth’s atmosphere.

This helps address a major contributor to space clutter. Most satellites are not designed to return to Earth at the end of their missions. Once they run out of fuel or stop working, they can drift in orbit for years or decades before gravity and atmospheric drag eventually bring them down. By slowing these satellites with an electrodynamic tether, Zhu’s system accelerates their orbital decay, helping them re‑enter the atmosphere far sooner than they would naturally.

Since 2010, he has been pursuing this technology as a way for satellites and spacecraft to be pre‑emptively equipped with disposal systems, allowing them to safely remove themselves at the end of their missions without adding new refuse or relying on costly clean up efforts. This approach could make sustainable orbital management the default, rather than the exception.

One of his projects, called DESCENT, put this concept into practice as Canada’s first on‑orbit test of space debris removal technology. Launched from the International Space Station, the Canadian Space Agency–funded mission consists of two CubeSat satellites connected by a 100‑metre electrodynamic tether, which will deploy in orbit to demonstrate how the system can actively lower a satellite’s orbit.

Complementing this work is Zhu’s research in autonomous space robotics, which he pursues alongside his efforts in keeping space clean. His lab develops systems capable of tracking, approaching and manipulating free‑floating and tumbling objects, using advanced perception, robotic dexterity, AI‑enabled decision‑making and control strategies to rendezvous with and grasp challenging targets. While these systems are developed primarily for on‑orbit servicing – such as repairing, refuelling or upgrading satellites without human spacewalks – Zhu believes they also have important applications for active debris removal, where autonomous robots can identify and capture defunct or tumbling objects in orbit.

Building on the autonomous robotic work, Zhu is exploring advanced swarm‑based approaches. He swarms of small satellites that autonomously coordinate to locate and interact with the waste. “My concept is very cheap, small satellites that can be mass‑produced, launched into space and then work as a swarm,” he says. “It’s decentralized control – more like ants. When one satellite finds a target, it shares the information so others can approach without collision among themselves and coordinate to dock onto or push the debris.”

Each satellite is designed to nudge or influence space litter using tethers or contact‑based mechanisms, rather than complex robotic arms, and the swarm is intended to deorbit along with the debris after interaction.

Currently, as part of his Tier 1 York Research Chair in Space Robotics and AI (2024-29) and as director of the NSERC CREATE Program in Smart Autonomous Robotic Technology for Space Exploration (SMARTART), Zhu is actively publishing and presenting on these concepts while nurturing the next generation of engineers and researchers who could bring them to fruition. Through SMARTART, students gain industry‑oriented training in AI, autonomous robotics, computer vision and systems engineering, equipping them with the skills needed to tackle challenges like coordinated spacecraft swarms and active debris removal.

Seeing his students embrace these ideas and contribute to the field, Zhu notes the growing global engagement with space debris issues.

As someone who once felt he was among the few raising concerns about space debris in 2010, Zhu is encouraged by the reception and interest his work now receives, as well as the efforts he sees worldwide from researchers and organizations.

“My reward is seeing more people following my path to do this,” he says. “I’m glad to see more people paying attention and recognizing the importance of this issue.”

The post �첥��Ƶ engineer develops solutions to make space more sustainable appeared first on YFile.

Since joining York in 2022, Oyibo noticed that students sometimes struggled to retain content between classes. This was especially true in memory-intensive disciplines like biology, psychology and human-computer interaction, which require knowing and application of key principles and concepts.

Oyibo wanted to find a way to help.

“I was trying to look for ways I can support students to learn and retain the content beyond the course,” he says.

He thought of mnemonics – memory aids that leverage how the brain encodes, stores and retrieves information. “Mnemonics can efficiently encapsulate knowledge in a way that makes it easier to recall and frees up cognitive resources for higher-order processes, such as understanding, analysis and synthesis,” he explains.

Oyibo drew on his expertise in machine learning, persuasive technology, human-centred design and creative writing. Persuasive technology involves designing digital tools that encourage users to take positive actions – such as fitness apps that motivate regular exercise or contact tracing applications that increase public health participation – without manipulation or deception. Human-centred design emphasizes building solutions around the needs, experiences and behaviours of real people, rather than forcing them to adapt to rigid systems.

Oyibo applied these principles to previous research, creating personalized digital health tools that adapt to different cultures and communities. This technology, he says, helps users stay healthy while addressing barriers that exclude underrepresented groups.

Now, he is focused on the development of the SANKOFA Project, a toolkit named after the Akan word “Sankofa” meaning “go back and fetch what is lost” and an acronym for "Save All New Knowledge Optimally and Fetch Accurately."

The toolkit leverages memory-enhancement principles and has two main application components: SAVE (Selection, Association, Visualization, Elaboration); and RADAR (Recollection, Association, Decoding, Artifacts Review). The SAVE tool allows students to create mnemonics using text, images, audio and video to encode complex information, while the RADAR app supports retrieval practice, helping learners recall, reflect on and reinforce learnings through interactive exercises and games.

Oyibo’s current work uses AI and genetic algorithms to optimize mnemonics for learning, aiming for what he calls “mnemonic singularity” – mnemonics that cannot be significantly improved in terms of effectiveness to maximize knowledge retention. He incorporates the SAVE tool and RADAR app in this work to promote engagement, active recall and consistent practice through interactive exercises and gamified learning.

The toolkit also accommodates different learning styles with multimodal mnemonics – text, images, audio and video – with plans to support translations to enhance accessibility across languages and cultures.

Early testing at York is encouraging and shows the SANKOFA Toolkit can improve learning and memory retention, says Oyibo. Pilot studies in Ghana are exploring how the approach generalizes in other educational contexts. His findings will be published over the next few months.

While the toolkit is currently designed to serve university students, Oyibo envisions scaling it to learners of all ages and deploying it far beyond York. “I want to organize the world’s knowledge using mnemonics – not just mnemonics, but effective mnemonics. I’m thinking of a platform where teachers and students can collaborate with gamification used to reward meaningful and useful contributions. The goal is deployment in classrooms, not just here in Canada, but globally.”

At its core, Oyibo’s work builds on the inspiration for the SANKOFA Project: helping people overcome barriers and achieve success.

This philosophy connects much of his past research. Before arriving at York, he worked at the University of Waterloo on contact tracing applications during the COVID19 pandemic, using persuasive design and personalized behavioural insights to improve public health engagement.

At York, he is focusing on inclusive design in fitness and health technologies, applying AI and machine learning to tailor digital tools for underrepresented populations – work that earned him a Gold Award at the 2019 Human-Computer Interaction International student design competition and a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada.

Across all these projects, Oyibo’s guiding principle remains the same: “I want to solve real-world problems that have a great impact and make meaningful contributions. I want to be a key player, not a spectator, in the global stage of research and development.”

The post York researcher creates AI tool to improve learning retention appeared first on YFile.

Over time, that gap takes on a second kind of damage: it wears away trust in the health care system. It's something �첥��Ƶ's Michael Kalu has seen play out in his research and heard directly from the communities he works with.

Part of what drives that erosion, he explains, is a feeling voiced by many Black older adults – fatigue from repeatedly raising the same concerns to clinicians and researchers, and seeing little change in return.

"They were exhausted with having to come back every single time to report the same needs over and over again," says Kalu, an assistant professor at York's .

That collective frustration became a guide for Kalu and helped shape the vision behind the Black-Focused Interactive Repository for Actionable Voices and Engagement (BiRAVE) – a Canadian Institutes of Health Research (CIHR)-funded project in its early stages.

Led by Kalu, the project aims to better understand health care needs of Black older adults – including access, affordability and quality – and will develop a tool to support reporting these health inequities. The CIHR funding supports interviews and the development of an initial prototype, with the team starting in Toronto to collect baseline information that will inform the first version of the platform.

The four-year project will expand to Winnipeg, Halifax and other provinces.

BiRAVE will function as a digital space where users can share experiences with gaps in health care – whether physical, environmental, cognitive or social – by using an AI chatbot.

"The chatbot will either be a part of an app or available on a website – it depends on what the older Black adults decide. It is important that the entire project is co-created with older Black adults," he says.

Kalu adds the team is exploring ways to make the tool accessible through typing or speaking, with offline access also being considered.

"In order to reflect the heterogeneity in the community, BiRAVE will also be culturally aware and include multiple languages," he says, adding the platform is intended to recognize accents and dialects, including Patwa and Pidgin English.

BiRAVE is planned around three connected parts: a space where Black older adults can report unmet needs; a system that suggests real-time solutions based on those reports; and a community forum where members can share experiences to help shape the platform over time. All reporting, he says, will be kept private.

"BiRAVE will work in a continuous loop. Over time, the project aims to build a clearer picture of where gaps in care persist and what solutions could help close them," says Kalu, noting that to his knowledge, BiRAVE would be the first "living" interactive repository of its kind in Canada.

The project brings together a Canada-wide team of about 30 Black researchers and five advisory committee members and partner principal investigators from other Ontario universities: Barbara Hamilton-Hinch, Lydia Kapiriri, Ojembe Blessing, Rita Orji, Salami Oluwabukola and Ingrid Waldron.

In addition to the CIHR project grant, BiRAVE has received funding from �첥��Ƶ's Faculty of Health Collaborative/Community Research Seed Grant, the Connected Minds EDI funding and the Canadian Association on Gerontology New Investigator Award.

The research team is also working with 17 community organizations, including Black Creek Community Health Centre, Rexdale Community Health Centre, TAIBU, Sisterhood and Brotherhood Nova Scotia and Afrimama Manitoba.

In the long run, Kalu says BiRAVE's repository will serve as a centralized resource for researchers, clinicians and policy leaders, providing real-time insight into the social and health needs of Black community members.

"Because BiRAVE will be organized by local areas, data from the platform could help health authorities address health inequities in their region," he says, highlighting the project's focus on addressing real-world issues through community-informed and innovative approaches.

With files from Mzwandile Poncana

The post York collaborates on project to advance equitable health care appeared first on YFile.

In recent years, AI tools have become widely adopted across a range of sectors, supporting many applications. Many of these tools are powered by deep learning (DL) models, which learn patterns directly from large volumes of data by adjusting internal parameters – the settings a model changes as it learns. An example of this is the large language model (LLM).

For some organizations, however, adoption can prove challenging. “One main challenge is the cost and time required to train billions of parameters from collected data, and the cost of using LLMs for interference, which generate text answers, images, voices and vidoes based on requests,” says Cheung. “Its sheer scale translates to large daily operation costs for businesses.”

There are also environmental considerations. The longer it takes to train an AI model, the more electricity is consumed. “This is why AI tech companies currently demand huge energy supplies and are building nuclear plants to meet their needs,” he says. Much of this electricity comes from carbon‑intensive sources, contributing to greenhouse gas emissions and environmental impact.

Over the last several years, Cheung, a faculty member in the Department of Electrical Engineering and Computer Science, has worked with international partners to develop smaller, more efficient alternatives.

His research focuses on miniaturizing transformer models, a type of deep learning model that excels at discovering complex patterns in data – from language and images to time‑series signals – by analyzing relationships across all parts of the input.

“If we can miniaturize DL models by one to two orders of magnitude, that would mean much smaller training and operation costs,” he says. “It would also save a lot of electricity and thus be environmentally friendly.”

To date, Cheung’s research group at York has applied miniaturized models to a variety of areas, including imaging, traffic and weather data, and even brain signals measured by EEG. They have already reduced model sizes by up to 100 times without noticeable drops in performance. Their smaller models can perform image processing tasks, such as denoising and interpolation, as effectively as much larger state‑of‑the‑art systems, while using only a fraction of the model parameters.

Recently, Cheung and his collaborators received an opportunity to explore a new application of their innovative methods: LLMs, some of the largest and most widely used AI systems today. These models are trained on massive collections of text to understand and generate human language.

Last summer, Cheung connected with Eugene Wen, vice-president and global chief data scientist at Manulife, a multinational insurance and financial services company investing in advancing its AI capabilities and applied AI research.

"Our goal is to build advanced AI solutions that balance high-speed and accuracy with low energy consumption to reduce costs and our carbon footprint," says Wen.

The company was seeking an LLM that could answer customer queries quickly and accurately while using minimal computing resources and electricity, helping to keep costs and energy use low. Manulife provided funding for Cheng's research, including support for PhD students to participate, continuing its commitment to parter with universities on joint research projects.

Now, in partnership with Manulife, Cheung is pursuing this project in collaboration with his graduate students and Professor Vicky Zhao, a longtime friend and research partner from Tsinghua University in China.

Building on their previous work, Cheung and his collaborators are exploring a novel approach to applying their miniaturization techniques to LLMs. In an earlier project, they trained a parameter-efficient graph-based denoiser – an AI system that gradually removes noise from grainy images to produce a clear result using a learned similarity graph.

Generating text from scratch in an LLM can also be interpreted as a sequence of denoising steps, so that the developed denoiser can be redeployed in the language context. By training the generative model as sequential denoising stages, they hope to reduce the number of parameters needed, speed up training and lower energy use. This could create smaller, faster and more efficient LLMs.

Cheung says the work with Manulife also allows him to pursue his broader research philosophy. “The main driver of my research is to understand,” he says.

He notes that most off‑the‑shelf LLMs operate like black boxes, with limited visibility into why different operations are stacked together in particular configurations. By applying his miniaturization techniques to LLMs, he can test these ideas on a new type of AI system, learning what the model truly needs to know and reducing unnecessary complexity.

“As signal processing researchers, my colleagues and I strive to understand systems in a more fundamental way so that we learn only what needs to be learned – the ‘known unknowns.’ In so doing, we reduce model parameters,” he says. This approach helps create smaller, more efficient language models and furthers his goal of understanding AI at a foundational level.

Manulife data scientist are part of the research team, providing data and experience in building generative AI solutions to solve business challenges. This arrangement allows Cheung and his team to continue research and refine models for real-world imipact, while Manulife can explore practical applications, such as reducing operational costs and environmental footprint. It also enables Cheung to pursue the broader objectives that drive his work.

“We hope that the long‑term impact of the research is to enable more frugal model learning that is more energy‑efficient and environmentally friendly,” he says. “That way everyone can benefit from the power of AI without paying a substantial environmental price.”

The post York prof innovating smaller, faster and more sustainable AI appeared first on YFile.

Using an AI-powered system called DePerio, the technology promises faster, more reliable diagnosis and improved periodontal care.

Patented through Innovation York, the breakthrough development in oral health care is advancing toward commercialization. Innovation York supports researchers in translating discoveries into real-world solutions.

Periodontal disease is a leading cause of tooth loss and a risk factor for systemic health conditions. Early detection is important, but current diagnostic methods produce inconsistent results due to a lack of sensitivity.

“DePerio uses advanced artificial intelligence to provide precise, data-driven assessments, offering clinicians a powerful tool to improve outcomes and reduce health care costs,” says Ebrahim Ghafar-Zadeh, who is the principal investigator and co-inventor of DePerio, as well as an associate professor in .

Through deep learning – a type of AI – DePerio works to analyze dental images and give exact measurements to help dentists make more informed treatment decisions through a user-friendly interface.

“By integrating deep learning into dental diagnostics, we can deliver faster, more accurate evaluations that benefit both patients and practitioners,” says Ghafar-Zadeh, noting York collaborated with researchers from University of Toronto for this project.

Findings from the research are published in the and explain how AI is used in DePerio to assess dental images and reduce errors in detecting disease.

An earlier study in focuses on the advanced algorithms behind the novel technology and how they integrate with intelligent systems for advanced health care. It also explores future applications of AI in oral health and suggests DePerio could be scaled for broader clinical use.

Building on this success, the team has secured an Idea to Innovation grant from the Natural Sciences and Engineering Research Council of Canada to advance the platform toward market readiness. This follows last year’s Connected Minds Prototyping Award, which supported early-stage development.

Several additional manuscripts and grant applications related to DePerio are currently under review, signaling strong momentum for the project.

“This recognition from NSERC and Connected Minds validates the potential impact of DePerio,” says Ghafar-Zadeh. “We are excited to continue refining the technology and exploring partnerships that will bring it into clinical practice.”

The post York researchers patent AI technology to transform dental care appeared first on YFile.