In collaboration with the Applied Perception Lab at Concordia University, research led by Dr. Amin Madani at The Institute for Education Research at UHN has developed LapBot—an innovative mobile game transforming how surgeons practice their skills for laparoscopic cholecystectomy, a common gallbladder removal surgery.
Developed as an educational tool, LapBot leverages artificial intelligence (AI) to provide instant feedback on safe surgical techniques, helping players sharpen their decision-making skills without stepping into an operating room.
In the game, players navigate real operative scenarios, identifying safe dissection points in the hepatocystic triangle, a critical area in gallbladder surgery. Using an AI-powered algorithm, the game scores players on their accuracy and offers personalized feedback on their choices.
A recent study involving 903 participants from 64 countries demonstrated LapBot’s effectiveness. Results showed that as the game increased in difficulty, players' scores and confidence levels dropped, reflecting the real-life challenges of surgical decision-making. Interestingly, higher scores were closely linked to players' surgical experience, affirming the game's ability to differentiate expertise levels.
Feedback from users was overwhelmingly positive as players noted support for LapBot’s integration into formal training programs, an improved ability to reflect on feedback during surgery, and enhanced learning when observing procedures.
LapBot exemplifies how mobile games can revolutionize medical education, making training more engaging and accessible. By transforming traditional learning into an interactive experience, it offers a promising new avenue for developing and refining surgical skills in a low-stakes, high-impact way.
This work was supported by UHN Foundation and the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) Foundation. Dr. Amin Madani is an Assistant Professor in the Department of Surgery at the University of Toronto and Surgical Artificial Intelligence Research Academy (SARA) at UHN.
Dr. Amin Madani is a consultant for Johnson & Johnson and is also on the Board of Directors for Global surgical Artificial Intelligence Collaborative.
A full list of author disclosures can be found on the article here.
St John A, Khalid MU, Masino C, Noroozi M, Alseidi A, Hashimoto DA, Altieri M, Serrot F, Kersten-Oertal M, Madani A. LapBot-Safe Chole: validation of an artificial intelligence-powered mobile game app to teach safe cholecystectomy. Surg Endosc. 2024 Sep. doi: 10.1007/s00464-024-11068-3.
Turning scientific discoveries into clinical applications can take many years of development and testing. That is why Commercialization at UHN has established the Accelerator Fund (AF), which provides financial and business development support to help promising technologies reach their full potential. Advancing research and clinical innovations through the AF is a joint effort between UHN researchers and business development experts from the Commercialization at UHN team, facilitating their market entry as new medical services, devices, or therapies.
The AF has two streams: the Innovation Accelerator and the future Clinical Accelerator. UHN Commercialization is thrilled to announce the first recipients of funding from the Innovation Accelerator stream.
Congratulations to the following UHN researchers who have received a total of $750,000 in internal funding to advance their work toward commercialization:
● Dr. Sarah Crome (Toronto General Hospital Research Institute): This funding will help support progress on a novel cell therapy to prevent or treat graft versus host disease following stem cell transplant to improve clinical outcomes.
● Dr. Shinichiro Ogawa (McEwen Stem Cell Institute): This funding will help in developing a stem cell-based, functionally complete, vascularized liver tissue for regenerative therapy in patients with liver failure.
● Dr. Pamela Ohashi (Princess Margaret Cancer Centre): Funding will go towards work on the development and pre-clinical validation of a novel cell therapy agent to be used for cancer immunotherapy.
Thanks to a generous partnership with UHN Foundation and The Princess Margaret Cancer Foundation, the Innovation Accelerator stream provides a maximum of $250,000 in funding per project—helping take promising novel technologies further towards commercialization and future patient impact.
Learn more about the winning projects here.
● Dr. Eleanor Fish is an Emerita Scientist at the Toronto General Hospital Research Institute. Dr. Fish studies antiviral treatments to stop the spread of viruses and lessen the severity of infections, with ongoing efforts to develop nasal sprays for preventing illnesses like the flu, RSV, and COVID-19.
● Dr. Masoom Haider is a Clinician Scientist at the Princess Margaret Cancer Centre. His work uses machine learning and AI to develop predictive tools for prostate and pancreatic cancer.
● Dr. Robert Inman is a Senior Scientist at the Krembil Research Institute. Dr. Inman investigates how the immune system and environmental factors contribute to autoimmune diseases and uses advanced techniques to understand ankylosing spondylitis.
● Dr. Jonathan Irish is a Senior Scientist at the Princess Margaret Cancer Centre. His research interests are health services, population-based outcomes in head and neck cancer, molecular techniques for predicting clinical outcomes, and quality-of-life in head and neck cancer.
● Dr. Milos Popovic is a Senior Scientist and Director at the KITE Research Institute. His work seeks to improve movement and rehabilitation through advanced technologies like electrical stimulation and robotics.
● Dr. Nicole Woods is a Senior Scientist and Director at The Institute for Education Research. Her research examines the role of basic science knowledge in clinical reasoning and the development of medical expertise.
● Dr. Brad Wouters is the Executive Vice President of Science and Research at UHN and a Senior Scientist at the Princess Margaret Cancer Centre. His research focuses on discovering ways to personalize cancer treatment to improve patient outcomes.
For the full list of 2024 CAHS Fellows, read the full press release here.
From virology to immunology
“My adventure in science started on the day I took a virology class in university. It completely changed my life.”
A business student at that time, Dr. David Brooks did not expect that the virology class would put him on a totally different track. It marked the start of his life-long research career on the immune system and its failure to fight long-term chronic diseases, from viral infections to cancers.
“I was fascinated by viruses and how a small, non-living entity—that sometimes only has four genes—could completely commandeer a cell for its own purposes.” David recalls with excitement.
After completing his PhD at UCLA on how HIV hides from human immune responses, David realized that studying viruses only tells half of the story. He needed to understand the immune system better to see why it is outsmarted in chronic infections.
A concept called “T cell exhaustion” caught his attention.
T cell exhaustion happens when CD8+ T cells—a type of immune cells that can recognize and destroy cells infected by viruses, other pathogens, as well as cancer cells—no longer respond effectively after a long period of stimulation from binding to these harmful cells.
“T cell exhaustion is a natural response that our body has evolved to stop our immune system after T cells have presumably completed their job. However, in the case of certain cancers or chronic viral conditions, the immune system stops responding even though its job is not yet done.”
David delved into T cell exhaustion during his postdoctoral studies in Dr. Michael Oldstone's lab at The Scripps Research Institute. The team used a preclinical model infected with Lymphocytic Choriomeningitis Virus (LCMV) to probe the immune responses to the virus.
In a study published in Nature Medicine, David discovered that blocking IL10—a small protein produced by immune cells that suppresses immune functions—can prevent CD8+ T cell exhaustion during viral infection and lead to clearance of the otherwise chronic infection.
“This work sparked my interest in how cellular factors can signal T cell exhaustion, laying the foundation for the rest of my career.” David comments in retrospective.
David established his own lab at UCLA and continued investigating how IL10 contributes to T cell exhaustion. One day, one of his postdoctoral students, Dr. Elizabeth Wilson, suggested: “Since type 1 interferon (IFN-I) can induce IL10, why not block interferon and see if it improves immunity against viruses?”
David was skeptical at first, as interferons are proteins known for disrupting viral replication and activating immune cells. “Removing an anti-viral system will not help in fighting a viral infection.” David predicted at that time.
Surprisingly, experiments showed that without IFN-I, the viral infection in LCMV models not only showed no signs of getting worse but it was actually eliminated within a few weeks.
“This was one of our most significant observations,” says David, “it showed a very different effect of interferons, completely contrary to what was known to the field.”
Through these studies, the team found that the immune system uses type 1 interferons to gauge how well it is fighting an infection. More interferons indicate more viruses present. When interferons are completely removed, this essentially gives immune cells an ‘inflammatory holiday’ and enables them to take a break from responding to inflammation signals. These cells can then regroup and revert into a state where they are better equipped to fight infection (published in Science).
T cell exhaustion is also seen in cancer.
“There are many cancers that behave like chronic viral infections,” David comments. “They constantly stimulate the immune system to respond, which can lead to T-cell exhaustion.”
David moved his lab to the Princess Margaret Cancer Centre in 2015 to study T cell exhaustion in human tumour models. Knowing that interferons can both drive and suppress immune responses, David’s next step was to determine their role in cancer.
“We found that while interferons can provide necessary anti-cancer signal on T cells, when this signal is prolonged, T cells become exhausted and lose their ability to fight tumours,” says David. “This switch to suppression is particularly devastating because it underlies the failure of multiple types of otherwise highly effective therapies.”
So how do the signals produced by interferons switch from good to bad?
Dr. Sabelo Lukhele, a Postdoctoral Researcher in David’s lab, looked at T cells in the tumor microenvironment and observed a spike of a protein called IRF2 in response to sustained interferon signaling. The team found that after interferons send signals to T cells, cellular protein IRF2 increases to regulate genes that dim the immune function of the cell. They shut down IRF2, which gave the immune system a boost right when it started to lag in the fight against cancer (published on Immunity).
“We found that IRF2 is the switch that turns positive responses in T cells to a negative, suppressive one,” says David excitedly. “Reversing this allowed T cells to retain their function and control tumour growth in pre-clinical models.”
To turn off this immunosuppressive switch, David’s team innovatively designed CAR T cells with the IRF2 protein removed to prolong the effectiveness of the therapy. CAR T cells are essentially engineered T cells with surface receptors that specifically target cancer cells, enhancing their anti-cancer capabilities.
“We think that removing IRF2 will enable CAR T cells to finish the job of killing the cancer,” says David. “This new approach has been patented and we are currently testing it in pre-clinical studies with the hope of moving it into the clinic in the future.”
In addition to developing new therapies, David’s research on interferon also sheds light on the effectiveness of existing cancer treatments.
When a patient’s T cells have a strong reaction to interferons, the patient is unlikely to respond well to anti-PD-1, a type of immunotherapy. On the other hand, those that react weakly to interferons have a better outcome for the therapy. David’s team published this observation in Nature Immunology, showing the predictive power of interferons.
“A key challenge in cancer treatment is predicting how patients will respond to specific therapies. This will allow us to find alternative options for those who do not respond well.” David adds.
David’s team is exploring the use of interferon signaling as a biomarker to predict how cancer patients will respond to immunotherapy. This innovative strategy, recently patented and under development at the Princess Margaret, aims to guide therapeutic decisions in clinical settings.
“We want to get the right therapy to the right people,” David asserts, expressing hope that this approach will make that possible.
“It was a humbling experience jumping from virology to immunology and from studying viral infection to studying cancer; there was so much to learn.”
For David, it is the intellectual stimulation and the opportunity to help cancer patients that drive him every day to advance in research.
“The questions we ask each day are new. The things we uncover each day are new,” says David. “One of the great things about science is that it offers endless opportunities to deepen our understanding of biology and leverage that knowledge to combat disease.”
Meet PMResearch is a story series that features Princess Margaret researchers. It showcases the research of world-class scientists, as well as their passions and interests in career and life—from hobbies and avocations to career trajectories and life philosophies. The researchers that we select are relevant to advocacy/awareness initiatives or have recently received awards or published papers. We are also showcasing the diversity of our staff in keeping with UHN themes and priorities.
Scientists at Toronto General Hospital Research Institute (TGHRI) have developed an improved method for evaluating the performance of Artificial Intelligence (AI) models across various health care settings.
As health care datasets become larger and more complex, the use of AI for the analysis of these datasets is gaining traction. Medical information can take the form of unstructured data such as medical images, electrocardiograms (ECGs), and text from clinical notes. Despite advancements in AI that have produced tools capable of analyzing medical images and clinical language, it remains challenging to predict their effectiveness in different health care settings without testing on new and varied data from each setting.
“For AI tools to be truly safe and effective for patient care, they must perform reliably across different situations and patient groups, a concept known as generalizability, which requires accurate performance estimation,” says Cathy Ong Ly, doctoral student at TGHRI and co-first author of the study. “We sought to address this challenge of estimating AI model accuracy by analyzing 13 datasets across different modalities such as X-rays, CT scans, ECGs, clinical notes, and lung sound recordings.”
When the team tested various AI models on this data, they found that their performance was often overestimated by about 20% on average. “We propose that this overestimation is due to data acquisition bias (DAB), a natural occurrence when data for these studies is retrospectively collected from regular medical care,” says Dr. Chris McIntosh, Scientist at TGHRI and senior author of the study.
“Generally speaking, AI might focus on irrelevant patterns in the data instead of what really matters for the task,” adds Dr. McIntosh. “Different hospital departments may use different equipment or settings and have different patient acquisition conditions. These variations, which might be imperceptible to researchers and clinicians, can be detected by AI algorithms. When models are trained on this data, they might rely on these subtle differences—like how a medical image was taken—rather than the actual medical content, to make predictions.”
An example of this bias is how patients suspected of having interstitial lung disease are often directed towards specific imaging techniques meant to confirm the diagnosis, while those without suspicion get more general scans. The algorithm will appear highly accurate at the hospital the data was trained on, but when deployed for clinical care at another hospital with different scanners, the accuracy will drop, potentially putting patients at risk.
To address this issue, the researchers developed and proposed an open-source accuracy estimate called PEst that corrects for bias and provides more accurate estimates of a model’s external performance.
“Our method, which corrects for hidden patterns and biases in the data, predicts models performance on new datasets with an accuracy margin within 4% of the actual results,” says Balagopal Unnikrishnan, doctoral student at TGHRI and co-first author of the study.
Given how crucial the accuracy of AI models is in health care, where recommendations can significantly impact patient outcomes, these findings will help enable safer and more widespread use of AI and support the development of new medical AI technology. This study was a truly multidisciplinary effort across UHN to measure the impact of these biases in a diverse array of modalities and diseases.
Cathy Ong Ly (Left) and Balagopal Unnikrishnan (middle) are doctoral students in Dr. Chris McIntosh’s lab (right).
This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), The Princess Margaret Cancer Foundation, and UHN Foundation. Data for this study was supported by foundation investments in the Digital Cardiovascular Health Platform including the Peter Munk Cardiac Centre and Ted Rogers Centre for Heart Research and MIRA through Cancer Digital Intelligence.
Dr. Chris McIntosh is an Assistant Professor in the Department of Medical Biophysics at the University of Toronto (U of T). He holds the Chair in Artificial Intelligence and Medical Imaging at the Joint Department of Medical Imaging at UHN and the Department of Medical Imaging at U of T.
Ong Ly C, Unnikrishnan B, Tadic T, Patel T, Duhamel J, Kandel S, Moayedi Y, Brudno M, Hope A, Ross H, McIntosh C. Shortcut learning in medical AI hinders generalization: method for estimating AI model generalization without external data. NPJ Digit Med. 2024 May 14;7(1):124. doi: 10.1038/s41746-024-01118-4.
A recent study led by McEwen Stem Cell Institute researchers at UHN found that simply using gene-edited stem cell-derived heart cells to replace damaged heart tissue was insufficient in preventing life-threatening arrhythmias. This work suggests that more complex approaches are needed to overcome critical irregular heartbeats.
Heart disease is a leading cause of death worldwide, with heart attacks often leaving lasting damage to the heart. Scientists have been exploring the use of human stem cell-derived heart cells, known as cardiomyocytes, to regenerate damaged heart tissue. However, a significant challenge has emerged as these transplanted cells sometimes trigger irregular heartbeats or arrhythmias.
To address this issue, a research team led by Dr. Michael Laflamme, Senior Scientist at the McEwen Stem Cell Institute and senior author of the study, targeted a specific regulator of the heart’s electrical behaviour, the HCN4 ion channel. This HCN4 protein provides heart muscle cells with pacemaker activity and is known to contribute to abnormal electrical activity in stem cell-derived heart muscle cells.
The team used advanced gene-editing tools to create modified heart cells with a less active HCN4 ion channel, aiming to reduce the risk of arrhythmias. However, upon transplantation into damaged heart cells, the researchers found that both the modified and unmodified cells still caused frequent arrhythmias.
“These results were surprising,” said Dr. Rocco Romagnuolo, who previously worked as a postdoctoral fellow in Dr. Michael Laflamme’s lab and is a co-first author of the study. “Despite evidence suggesting that the modification helped reduce the activity of this ion channel in the transplanted heart muscle cells, this strategy alone was not enough to prevent arrhythmias.”
Dr. Fanny Wulkan, a postdoctoral fellow in Dr. Laflamme’s lab and co-first author of the study adds, “This study highlights the critical reality of cardiac regeneration research: while it looks promising in the lab, translating these results to real-world therapies presents significant challenges. We need to better understand how these transplanted cells interact with the heart and target the factors that may contribute to the arrhythmia risk.”
As researchers continue to push the boundaries of regenerative medicine, studies like this are crucial to ensuring that new treatments are not only effective but also safe for patients. Future work will focus on identifying additional strategies to enhance the safety of these therapies.
This work was supported by UHN Foundation, the Canada Research Chairs Program, the University of Toronto’s Medicine by Design, and BlueRock Therapeutics. Dr. Michael Laflamme is a Professor in the Department of Laboratory Medicine and Pathobiology at the University of Toronto. Dr. Laflamme also holds a Canada Research Chair in Cardiovascular Regenerative Medicine.
Dr. Michael Laflamme is a scientific founder and paid consultant for BlueRock Therapeutics (BRT). Rocco Romangnuolo is a current BRT employee and Kyung-Phil Kim was a previous BRT employee.
#Wulkan F, #Romagnuolo R, Qiang B, Valdman Sadikov T, Kim KP, Quesnel E, Jiang W, Andharia N, Weyers JJ, Ghugre NR, Ozcan B, Alibhai FJ, Laflamme MA. Stem cell-derived cardiomyocytes expressing a dominant negative pacemaker HCN4 channel do not reduce the risk of graft-related arrhythmias. Front Cardiovasc Med. 2024 Jul 9. doi: 10.3389/fcvm.2024.1374881.
# Both authors contributed equally.
The current tools used to predict which children with pneumonia might become seriously ill or face life-threatening situations are not always reliable or objective. Scientists at UHN’s Toronto General Hospital Research Institute (TGHRI) have uncovered a potential way to identify children at risk of severe and fatal pneumonia.
“Pneumonia is one of the top causes of death in children, leading to over 700,000 deaths each year worldwide,” says Dr. Kevin Kain, Senior Scientist at TGHRI and co-senior author of the study. “This high mortality rate is even worse in low-resource environments, where other risk factors like malnourishment, coexisting infections such as human immunodeficiency virus (HIV), overcrowding, lack of immunization or vaccine availability, and indoor air pollution also exist.”
Currently, it is hard to reliably identify children with pneumonia who are at risk of a severe or life-threatening episode using existing tools. Scientists have recently started studying a protein called heparin-binding protein (HBP) as a possible marker to identify severe pneumonia and sepsis, but there is still not enough information about how reliable it is.
HBP is a protein that the immune system releases in response to infection. Research suggests that HBP might be involved in causing the lungs to fill with fluid during pneumonia. However, no previous studies have evaluated the potential for HBP to predict in-hospital mortality in children presenting with pneumonia.
“We sought to determine whether HBP measured at presentation to hospital could help identify Ugandan children at risk of a fatal outcome and improve the accuracy of medical assessments that determine which children need urgent care,” says Dr. Hridesh Mishra, Postdoctoral Researcher in Dr. Kain’s lab and first author of the study. “To achieve this, we studied 778 Ugandan children under 5 years old who were diagnosed with pneumonia.”
The study found that in patients with respiratory distress, the HBP concentration increased with increasing disease severity. In addition, children with higher HBP levels at admission were more likely to have fatal outcomes compared to those with lower levels. Specifically, children with HBP levels above 41 ng/mL were over five times more likely to have fatal outcomes compared to those with lower levels.
The team also used a method called ROC curve analysis to determine if including HBP measurements could improve existing clinical scores—like the RISC (The Respiratory Index of Severity in Children) score—in predicting which children with pneumonia are at risk of fatal outcomes. The ROC curve is a tool used to evaluate how accurately a test or model can distinguish between different outcomes, such as survival versus fatality.
“For children with pneumonia, a higher RISC score indicates a greater need for immediate medical attention,” adds Dr. Mishra. “In our study, the RISC score alone was already effective at predicting the risk of death. However, combining it with HBP measurements made it even more accurate.”
The results of this study suggest that HBP measurements may help save lives by identifying those who need more intensive care.
This study was a collaborative effort with Dr. Sophie Namasopo from the Department of Paediatrics, Kabale Regional Referral Hospital, Kabale, Uganda and Dr. Michael T. Hawkes, Investigator and Pediatric Infectious Diseases Consultant at BC Children's Hospital serving as co-senior authors.
This work was supported by the Canadian Institutes of Health Research (CIHR), Canada Research Chairs program, Government of Spain’s Ministry of Science, Innovation, and Universities Generalitat de Catalunya, Spanish Ministry of Universities, and UHN Foundation.
Dr. Kevin Kain is a Professor in the Department of Medicine at the University of Toronto.
Mishra H, Balanza N, Francis C, Zhong K, Wright J, Conroy AL, Opoka RO, Bassat Q, Namasopo S, Kain KC, Hawkes MT. Heparin-Binding Protein Stratifies Mortality Risk Among Ugandan Children Hospitalized With Respiratory Distress. Open Forum Infect Dis. 2024 Jul 8;11(7):ofae386. doi: 10.1093/ofid/ofae386.
(L-R) Dr. Hridesh Mishra, first author of the study; Dr. Kevin Kain, co-senior author of the study.
Research conducted at UHN's research institutes spans the full spectrum of diseases and disciplines, including cancer, cardiovascular sciences, transplantation, neural and sensory sciences, musculoskeletal health, rehabilitation sciences, and community and population health.
Learn more about our institutes by clicking below:
