Updates and highlights from fall at the Krebil Centre For Neuroinformatics
No images? Click here
David Rotenberg
Chief Analytics Officer
Director of Operations, Krembil Centre for Neuroinformatics (KCNI), CAMH
Email Us:
Krembil.Centre@camh.ca for questions, feedback and to learn more about the KCNI!
A message from Director:
This fall has been a season of convergence at KCNI.
Our September Open House brought together scientists, trainees, and collaborators to trace the full arc of our work—from genomic discovery to clinical implementation. What struck me most wasn't just the sophistication of individual projects, but how naturally they connect: predictive models informed by circuit biology, digital tools shaped by equity research, AI systems designed around human needs rather than replacing human judgment.
This is the essence of what makes KCNI distinctive. We don't just build better tools; we reimagine how knowledge flows between domains. From isoform atlases uncovering hidden causes of autism, to precision approaches for treatment-resistant depression, to ensuring our AI systems amplify equity rather than bias—each advance strengthens the connections between others.
The Brain Health Data Challenge Platform, now in active development with Brain Canada's support, embodies this philosophy at scale. We're creating infrastructure that positions Canada to lead in ethical, collaborative AI for brain health—not through centralized control, but by enabling researchers worldwide to build on each other's work while respecting privacy and governance.
As we close 2025, I'm grateful for a team that understands that transforming mental health care requires more than technological innovation. It requires us to work at the connections—between disciplines, between discovery and application, between what we build and who we build it for.
Featured Highlights
KCNI Open House
Each year we curate a public event that highlights the work of our scientists and trainees, while also opening our doors to you. On Monday, September 29, we hosted our 2025 Open House, bringing together scientists, trainees, collaborators and guests to explore how computational neuroscience, predictive analytics, and digital innovation are transforming mental health research and care.
The event traced a journey from biological discovery to clinical application through a series of panels and discussions.
We invited trainees to step on stage with our lab leaders and scientists to discuss key discoveries and trends across technology and mental health.
The first panel, moderated by Dr. Andreea Diaconescu, examined advances that allow researchers to witness and predict the biological expressions of mental illness. Drs. Shreejoy Tripathy, Etay Hay, and John Griffiths described how genomic data, circuit modeling, and whole-brain simulations are revealing the links between brain biology and treatment response. Trainees Milad Soltanzadeh and Sorenza Bastiaens shared research on treatment-resistant depression and brain rhythm modeling. Together, the group emphasized how computational tools can help identify risk earlier, develop less invasive assessments, and personalize care.
The second discussion, moderated by Dr. Marta Maslej, focused on predictive analytics and how embedded technology is reshaping the study and treatment of mental illness. Drs. Dan Felsky, Erin Dickie, Laura Sikstrom, and Tristan Glatard, with trainee Dr. Mohamed Abdelhack, explored how artificial intelligence (AI) is used to develop digital biomarkers and predictive models. Speakers stressed the importance of ethical, transparent, and fair AI systems that enhance—rather than replace—the human connections essential to mental health care.
The final session, moderated by Dr. Glatard, turned to clinical and digital interventions. Drs. Andreea Diaconescu, Gillian Strudwick, Marta Maslej, and Nelson Shen, with trainees Pamina Laessing and Karishini Ramamoorthi, discussed how digital tools—from AI chatbots to wearable sensors—are transforming care delivery. Panelists emphasized trust, usability, and equity in technology design and implementation, and reflected on how predictive models could one day help clinicians deliver more responsive and personalized treatment.
Across all three discussions, a common theme emerged: the future of mental health research depends not only on scientific innovation but on collaboration, ethics, and a commitment to keeping people at the centre. From modeling molecular pathways to implementing digital systems in clinics, KCNI researchers are bridging the gap between discovery and real-world impact.
The Brain Health Data Challenge: Building Ethical, Equitable AI for Brain Health
With support from a $4.9-million investment from Brain Canada, Dr. Tristan Glatard is leading CAMH in creating an ambitious initiative to bring ethical artificial intelligence (AI) to the forefront of brain health care. The Brain Health Data Challenge Platform (BHDC), launched in June 2025, will empower researchers across Canada and around the world to create AI models that can meaningfully improve care for people living with serious mental illness and neurological disorders.
This project is driven by Dr. Glatard but is rooted in the KCNI commitment to Open Science and collaboration. The work of establishing the BHDC represents the commitment of over 30 Co-investigators across six key sites (CAMH, UHN, T-CAIREM, McGill, CMI, and CNRS).
At their core competitive AI challenges unify research communities and support the creation of powerful predictive models of pathology but remain underutilized in Canadian datasets due to strict privacy rules. This project will drive AI innovation in mental and brain health research by supporting advances in disease understanding, treatment development, and personalized care.
In its first few months, the team has laid critical groundwork: launching a project website and communications hub, establishing legal agreements and governance structures across partner institutions, and initiating platform design and staffing. Four major AI challenges are now being planned for launch in 2026, focused on schizophrenia, depression, Parkinson’s disease and addictions.
Group photo from the public announcement of Brain Canada's landmark grants. Back row: Sarah Downey, Dr. Tristan Glatard; Front row: Drs. Aristotle Voineskos, Viviane Poupon, and Ali Khan
BAARD: Personalizing Care for Depression in Older Adults
Depression in older adults is a major public health concern, and many individuals do not respond to their first antidepressant. In fact, fewer than 20 per cent of older adults achieve full remission.
In September 2024 l Dr. Dan Felsky, and partners at Washington University in St. Louis and the University of Pittsburgh received a transformative $9-million research grant from the U.S. National Institutes of Health; to improve outcomes for older adults facing treatment-resistant depression.
Previous large-scale clinical trials have shown that adding medications like aripiprazole or bupropion can improve outcomes (29 percent) but results still vary widely between individuals. Leaving clinicians guessing which treatment might work best for which patient, and precious time lost for those seniors, who spend years of cycling through treatments that don’t work and symptoms that never fully lift.
The Biotype-assigned Augmentation Approach in Resistant Late-Life Depression (BAARD) study aims to end that trial-and-error experience. Led by CAMH experts, the team is building a clinical decision support tool to guide more personalized, data-driven treatment choices for seniors who have not responded to standard care.
In its first year, the team used data from a large multi-site trial (OPTIMUM) and its embedded biomarker study (OPTIMUM-Neuro) to develop predictive models. These models combined clinical, cognitive and brain imaging data to forecast how a person might respond to a second-line medication. The best-performing models exceeded predefined accuracy benchmarks—suggesting real potential to improve remission rates by matching treatments to individual profiles. Machine learning played a key role in comparing predictive approaches. When the models were used to simulate real-world treatment scenarios, older adults identified as “biomarker-positive had significantly better outcomes than those not predicted to benefit.
The next step: validate these tools in independent patient datasets. Ultimately, this work is about giving people faster relief and restoring hope when depression has taken so much away. If successful, BAARD could transform how depression is treated in older adults—turning guesswork into precision care and helping more people recover sooner.
Using AI to Advance Equitable Care in Psychiatry
Aggression and Violence in psychiatric emergency settings can endanger both patients and staff, and the current practices used to manage risk— such as seclusion or restraints—can themselves be traumatizing.
Clinicians rely on structured rating scales like the Dynamic Appraisal of Situational Aggression (DASA) to spot early warning signs; however, false positives are common. Up to half of patients flagged as “high risk” by DASA never go on to exhibit aggression. Meaning people are being subjected to unnecessary and harmful interventions based on their DASA score vs. their actual interactions with clinicians.
Artificial intelligence is starting to be introduced as a way to improve these assessments, but CAMH research led by Drs. Laura Sikstrom and Marta Maslej has shown that machine learning models can worsen existing inequities for patients from racially marginalized groups and those experiencing unstable housing. Their work shows that without intervention, these AI tools risk amplifying systemic biases, and push the most vulnerable people further into cycles of stigma, surveillance and trauma.
“While it is important to assess violence risk in mental health settings to support early intervention and avoid coercive approaches, it is concerning that racially and structurally marginalized individuals are more likely to be mistakenly classified as high risk,” says Dr. Sikstrom. “We need a paradigm shift, reframing risk as a reflection of unmet needs.”
Over the past year, the team published their findings in Psychiatric Services a monthly and received a CIHR Rising Star Award of $300,000 to launch the next phase of research. This new project, led by Drs. Sikstrom and Maslej together with fellow CAMH scientist Dr. Juveria Zaheer, will co-design a humane AI system called FARE+.
The tool does not label individuals as risky, instead identifying the systemic conditions that produce false positive predictions of aggression. The goal is to give clinicians better, fairer tools—so that they can intervene earlier, avoid overreacting, and create safer environments for everyone.
“The FARE+ represents not just a tool, but a commitment to care that is fair, inclusive and centred on dignity,” Dr. Sikstrom explains. “It’s about more than managing aggression—it’s about fostering compassion and equity to support recovery for all.”
KCNI Scientist Highlight
Dr. Shreejoy Tripathy
How identifying variations in isoforms might help us uncover causes of Autism Spectrum Disorder
Autism Spectrum Disorder (ASD) affects many individuals and families, yet the genetic drivers remain unknown in about 80% of cases. This gap exists because our current "maps" of human genes are incomplete. This is especially true of the developing brain. Genes can be spliced into multiple forms, called isoforms, which create different proteins. Standard databases often miss these variations. The mutations within these regions remain unexamined, potentially overlooking causes ASD.
Our team of cutting-edge experimentalists, geneticists, and neuroinformatics experts is utilizing revolutionizing technologies to uncover these hidden genetic factors. We integrated the largest long-read RNA sequencing dataset to date ever collected from human neurons and advanced proteomics to create the most detailed atlas of gene expression during human neuron development. Our preliminary data reveals a vast new landscape: we identified nearly 200,000 isoforms, over half of which were previously unknown. Crucially, we have strong evidence that many of these novel isoforms are functional and produce proteins essential for healthy brain development.
Building on this breakthrough, our proposed research recently funded by Brain Canada aims to confirm
the biological relevance of these discoveries and apply them to ASD genetics. First, we will systematically validate which of these novel isoforms are translated into functional proteins Second, we will leverage this atlas to re-interpret whole-genome sequencing data from almost 100,000 individuals with ASD and their unaffected siblings. We hypothesize that mutations previously dismissed as harmless may actually disrupt these newly identified proteins. By integrating these complex datasets, we aim to pinpoint these "hidden" mutations that may in fact be causing ASD in these individuals.
This work promises to significantly improve the genetic diagnosis of ASD, offering long-awaited answers to families and laying the essential foundation for personalized therapeutic strategies.
Dr. Shreejoy Tripathy
Scientist, at the KCNI Computational Genomic Modelling
Group photo of Triplab at the annual KCNI Halloween lunch.
TRAINEE HIGHLIGHT
Trainee - Nuo Xu - PhD Student
Nuo Xu received her MSci in Pharmacology from University College London in 2021. During her time in London, she initially aspired to be an experimentalist, studying experimental techniques such as intracellular electrophysiology. However, the COVID-19 pandemic required her to pivot to a computational project for her final year.This experience sparked her passion for biological data analysis. She is currently completing her PhD at the Department of Physiology at UofT, working in the Computational Genomics Lab led by Dr. Shreejoy Tripathy. Her research focuses on alternative splicing in neurodevelopmental disorders. While this link is well-established, mechanistic insights have been limited by the constraints of second-generation sequencing technology. To address this, her project utilizes third-generation sequencing, which overcomes these length limitations and allows her to obtain a more comprehensive view of the neurodevelopmental transcriptome. Beyond the biological questions, Nuo is also passionate about developing efficient, reproducible, and open-source bioinformatics pipelines.In her free time, Nuo has a healthy obsession with music. She loves discovering new music and going to concerts with friends. She can often be spotted wearing t-shirts acquired after waiting hours in line at the merch table.
Nuo Xu -PhD Student, MSci
Computational Genomic Modelling Lab - Dr. Shreejoy Tripathy
Congratulations to our recent Graduates from 2025!
We'd like to take a moment to acknowledge the significant work and contributions of Trainees who have recently Graduated or Defended their PhDs
Dr. Sorenza Bastiaens - Aug-2025 - PhD - IMS - Supervisor John Griffiths
Dr. Kevin Kadak - Sep-2025 - PhD - IMS - Supervisor John Griffiths
Dr. Earvin Tio - Sep-2025 - PhD- IMS - Supervisor Daniel Felsky
From left to right Drs. Kevin Kadak, Sorenza Bastiaens, and Earvin Tio at their PhD graduation ceremony
Featured Publications
- Faraz Moghbel, Muhammad Taaha Hassan, Alexandre Guet-McCreight, Heng Kang Yao, Etay Hay. Deriving connectivity from spiking activity in detailed models of large-scale cortical microcircuits. bioRxiv 2024.06.13.598937; doi: https://doi.org/10.1101/2024.06.13.598937
- Tio, Earvin et al. Understanding the Biopsychosocial Mechanisms of Risk for Suicide Using Machine Learning and a Resilience Framework. Biological Psychiatry, Volume 97, Issue 9, S356 - S357
Bastiaens SP, Momi D, Griffiths JD (2025) A comprehensive investigation of intracortical and corticothalamic models of the alpha rhythm. PLoS Comput Biol 21(4): e1012926. https://doi.org/10.1371/journal.pcbi.1012926
Momi, D., Wang, Z., Parmigiani, S. et al. Stimulation mapping and whole-brain modeling reveal gradients of excitability and recurrence in cortical networks. Nat Commun 16, 3222 (2025). https://doi.org/10.1038/s41467-025-58187-6
Griffiths JD, Kadak K. Towards a mathematical theory of rTMS-induced neural plasticity. Clin Neurophysiol. 2025 Sep;177:2110826. doi: 10.1016/j.clinph.2025.2110826. Epub 2025 Jul 1. PMID: 40675036.
About Us
The Krembil Centre for Neuroinformatics collaborates globally to collect and integrate large-scale brain research data, apply machine learning and artificial intelligence, and develop multiscale computational models that can transform our understanding of brain disorders. Our open, team science approach focuses on bridging the levels of the brain, from genes to circuits and from whole brains to the whole person, in order to better define, prevent and treat mental illnesses.
Learn more at
www.krembilneuroinformatics.ca
Want to Join Our Team? KCNI is always looking for scientists, post-doctoral fellows, grad students, coordinators, and more!
#KrembilNeuroinformatics is putting today’s most advanced technology to work on this universal task that will unlock the power of personalized medicine to change the world.
