The Bhat Transplant AI Lab develops and deploys next-generation artificial intelligence systems to improve access to liver transplantation, optimize organ allocation, personalize post-transplant care, and ultimately improve the lives of patients with liver disease.
Based at the Ajmera Transplant Centre, University Health Network and the University of Toronto, our multidisciplinary team of clinicians, computer scientists, engineers, and trainees develops AI solutions that address some of the most complex decisions in medicine. Our work spans the full liver transplant journey, from referral and waitlisting to organ allocation, transplant selection, graft monitoring, and long-term outcomes.
Our research has led to pioneering contributions in longitudinal deep learning, multimodal foundation models, and multi-agent AI systems for transplantation, with publications in leading journals including The Lancet Digital Health, Nature Communications, Journal of Hepatology, and JAMA Surgery.
Our innovations are now progressing beyond retrospective studies into prospective deployment and international clinical evaluation.
Our Research Pillars
AI for Equitable Access to Liver Transplantation
Goal: Ensure every patient has equitable access to life-saving transplantation.
Current initiatives include:
- Dynamic prediction of waitlist mortality and transplant benefit
- AI-driven identification of inequities in transplant access
- Fairness-aware allocation models
- Population-scale analyses using provincial and national datasets
- Public engagement and ethical frameworks for AI-assisted allocation
Thousands of patients die waiting for liver transplantation each year. We develop AI systems that identify disparities in access, predict waitlist mortality, and support more equitable referral and prioritization strategies.
AI-Augmented Transplant Decision Making
Goal: Build trustworthy AI systems that augment—not replace—clinical expertise.
Current initiatives include:
- Multi-agent AI transplant selection committees
- Explainable AI for high-stakes clinical decisions
- Human-AI collaboration frameworks
- International prospective deployment studies
- Governance and safety evaluation of clinical AI
Liver transplant candidacy decisions are among the most complex decisions in medicine, requiring multidisciplinary evaluation of clinical, psychosocial, and ethical factors.
Our group developed one of the first multi-agent AI transplant selection committee frameworks, in which specialized AI agents emulate the expertise of hepatologists, surgeons, coordinators, social workers, and ethicists to generate transparent, evidence-based recommendations. In international validation studies, these systems demonstrated high concordance with expert transplant committees and are now undergoing prospective evaluation.
Precision Monitoring of the Liver Allograft
Goal: Replace reactive care with proactive, personalized graft monitoring.
Current initiatives include:
- Deep learning prediction of graft fibrosis
- Multimodal pathology prediction
- Radiomics and imaging-based graft assessment
- Digital biomarkers of graft health
- AI-guided surveillance strategies
Long-term graft injury remains a leading cause of morbidity and graft loss following liver transplantation.
Our laboratory pioneered longitudinal deep learning approaches for non-invasive monitoring of graft health and developed GraftIQ, a hybrid neural network platform that integrates clinician expertise with machine learning to identify causes of graft injury without biopsy. These models have undergone international validation across North America, Europe, and Asia.
Predicting Long-Term Outcomes After Transplantation
Goal: Extend graft survival and improve quality of life for transplant recipients.
Current initiatives include:
- Dynamic prediction of graft failure
- Cardiovascular and metabolic risk prediction
- Dementia and neurologic outcomes after transplantation
- Reinforcement learning for immunosuppression optimization
- Clinical decision support systems embedded within electronic health records
Liver transplant recipients face lifelong risks of cardiovascular disease, malignancy, infection, chronic kidney disease, and graft failure.
We develop longitudinal AI models capable of integrating years of clinical data to generate individualized risk predictions and identify modifiable drivers of adverse outcomes. Our work leverages some of the world’s largest transplant datasets and focuses on real-world deployment of prediction models into clinical workflows.
Building the Future: Foundation Models and Agentic AI for Transplantation
Goal: Create intelligent AI ecosystems that support patients and clinicians across the entire transplant journey.
Healthcare AI is rapidly evolving from isolated prediction models toward intelligent systems capable of reasoning, planning, and collaborating.
Our laboratory is helping define the next generation of transplantation AI through development of multimodal foundation models and agentic AI systems capable of integrating clinical records, imaging, pathology, laboratory data, and expert knowledge.
Through national and international collaborations, we are establishing the infrastructure needed to safely evaluate and deploy these technologies across healthcare systems.
Training the Next Generation of Leaders
Our laboratory provides a highly collaborative environment for trainees interested in artificial intelligence, transplantation, clinical epidemiology, computational medicine, and implementation science.
Trainees gain experience in:
• Machine learning and deep learning
• Large language models and agentic AI
• Clinical trial design and AI deployment
• Health equity and responsible AI
• Translational research from development to implementation
We are committed to training future leaders capable of bridging medicine, data science, and healthcare innovation.
