vanderschaarlab/mlforhealthlabpub

Covers heterogeneous treatment effect estimation, survival analysis with competing risks, and synthetic data generation using GANs with differential privacy guarantees. Implements specialized architectures including multi-task Gaussian processes, deep kernel learning, and recurrent neural networks for temporal clinical data and time-series modeling. Spans causal inference, feature selection, uncertainty quantification, and counterfactual reasoning across diverse healthcare applications from prognostic modeling to organ transplant optimization.

463 stars. No commits in the last 6 months.