AI & Explainable Machine Learning
Predictive modeling, interpretable machine learning and explanation methods that help researchers understand what a model learned, where it may fail and how predictions relate to clinical context.
Research
An integrated research program from data to clinical evidence.
AI4Health connects heterogeneous health data and evidence with scalable computation and rigorous epidemiologic methods. We organize the program as a connected evidence system rather than a collection of isolated topics.
Research architecture
Three layers, one connected evidence system.
The same computational infrastructure can support multiple clinical questions, while the same clinical question can be examined across multiple evidence sources.
Layer 01Data & Evidence
Administrative health dataEHR & laboratory dataFAERS & Canada VigilanceDrug product monographsClinical trialsRegulatory documents
Layer 02Methods & Infrastructure
Machine learningExplainable AINLP & ontologiesCausal inferenceHigh-throughput computingAutomationVisual analyticsPrivacy-aware computation
Layer 03Clinical & Translational Applications
Medication safetyDrug-drug interactionsCKD & high-risk populationsClinical decision supportTrial generalizabilityPost-market surveillance
Core research areas
Where we concentrate our expertise.
High-Throughput Computing & Automation
Automated cohort generation, scalable computation and reusable pipelines that allow rigorous analyses to be repeated across many drugs, interactions, outcomes and populations.
Drug Safety & Real-World Evidence
Population-based studies of medication safety and effectiveness using active-comparator designs, confounding control, sensitivity analyses and clinically meaningful outcomes.
Pharmacovigilance & Safety Signals
Spontaneous-reporting systems, disproportionality analysis, terminology harmonization, missingness assessment and cross-source evidence to strengthen post-market safety surveillance.
Drug Monographs & Regulatory Intelligence
NLP, MedDRA mapping, ontology-driven extraction and temporal or version-aware analysis of evolving drug monographs, labels and regulatory knowledge.
Visual Analytics & Research Software
Human-centred research software for high-dimensional EHRs, multimorbidity, feature-group discovery, laboratory-based prediction and pharmacovigilance data quality.
Trials, Generalizability & Causal Methods
Assessing who is represented in trials, where evidence may not generalize, and how causal methods and target-trial emulation can strengthen treatment comparisons in routine care.
Reproducibility & Privacy-Aware Analytics
Provenance-aware workflows, standardized data processing, secure distributed analysis and computational practices that make large-scale health-data research more reproducible.
Interactive research explorer
Build an AI4Health research pathway.
Choose an evidence source, a method and an application. The explorer shows the research program that best connects them.
AI4Health pathway
Computational pharmacovigilance
Combining spontaneous-reporting evidence with interpretable computational methods can help prioritize safety questions, characterize reporting patterns and expose data-quality limitations before downstream inference.
Explore related projects →