Note : For the most recent papers and code, checkout https://github.com/vanderschaarlab.
Legacy code : This repository contains the implementations of algorithms developed by the van der Schaar Lab for papers before 2023.
An overview of the content of this repository is as below:
.
├── alg/ # Directory contains algorithms.
├── app/ # Directory contains apps.
├── cfg/ # Directory contains common config.
├── doc/ # Directory contains common docs.
├── init/ # Directory contains algorithms.
├── template/ # Directory contains templates.
└── util/ # Directory contains common utilities.The publications and the corresponding locations in the repo are listed below:
| Paper [Link] | Journal/Conference | Code |
|---|---|---|
| Bayesian Inference of Individualized Treatment Effects using Multi-task Gaussian Processes [Link] | NIPS 2017 | alg/causal_multitask_gaussian_processes_ite |
| Deep Multi-task Gaussian Processes for Survival Analysis with Competing Risks [Link] | NIPS 2017 | alg/dgp_survival |
| AutoPrognosis: Automated Clinical Prognostic Modeling via Bayesian Optimization with Structured Kernel Learning [Link] | ICML 2018 | alg/autoprognosis |
| Limits of Estimating Heterogeneous Treatment Effects: Guidelines for Practical Algorithm Design [Link] | ICML 2018 | alg/causal_multitask_gaussian_processes_ite |
| GAIN: Missing Data Imputation using Generative Adversarial Nets [Link] | ICML 2018 | alg/gain |
| RadialGAN: Leveraging multiple datasets to improve target-specific predictive models using Generative Adversarial Networks [Link] | ICML 2018 | alg/RadialGAN |
| GANITE: Estimation of Individualized Treatment Effects using Generative Adversarial Nets [Link] | ICLR 2018 | alg/ganite |
| Deep Sensing: Active Sensing using Multi-directional Recurrent Neural Networks [Link] | ICLR 2018 | alg/DeepSensing (MRNN) |
| DeepHit: A Deep Learning Approach to Survival Analysis with Competing Risks [Link] | AAAI 2018 | alg/deephit |
| INVASE: Instance-wise Variable Selection using Neural Networks [Link] | ICLR 2019 | alg/invase |
| PATE-GAN: Generating Synthetic Data with Differential Privacy Guarantees [Link] | ICLR 2019 | alg/pategan |
| KnockoffGAN: Generating Knockoffs for Feature Selection using Generative Adversarial Networks [Link] | ICLR 2019 | alg/knockoffgan |
| ASAC: Active Sensing using Actor-Critic Models [Link] | MLHC 2019 | alg/asac |
| Demystifying Black-box Models with Symbolic Metamodels [Link] | NeurIPS 2019 | alg/symbolic_metamodeling |
| Differentially Private Bagging: Improved Utility and Cheaper Privacy than Subsample-and-Aggregate [Link] | NeurIPS 2019 | alg/dpbag |
| Time-series Generative Adversarial Networks [Link] | NeurIPS 2019 | alg/timegan |
| Attentive State-Space Modeling of Disease Progression [Link] | NeurIPS 2019 | alg/attentivess |
| Conditional Independence Testing using Generative Adversarial Networks [Link] | NeurIPS 2019 | alg/gcit |
| Dynamic-DeepHit: A Deep Learning Approach for Dynamic Survival Analysis with Competing Risks based on Longitudinal Data [Link] | IEEE | alg/dynamic_deephit |
| Temporal Quilting for Survival Analysis [Link] | AISTATS 2019 | alg/survivalquilts |
| Estimating Counterfactual Treatment Outcomes over Time through Adversarially Balanced Representations [Link] | ICLR 2020 | alg/counterfactual_recurrent_network |
| Contextual Constrained Learning for Dose-Finding Clinical Trials [Link] | AISTATS 2020 | alg/c3t_budgets |
| Learning Overlapping Representations for the Estimation of Individualized Treatment Effects [Link] | AISTATS 2020 | alg/dklite |
| Learning Dynamic and Personalized Comorbidity Networks from Event Data using Deep Diffusion Processes [Link] | AISTATS 2020 | alg/dynamic_disease_network_ddp |
| Stepwise Model Selection for Sequence Prediction via Deep Kernel Learning [Link] | AISTATS 2020 | alg/smsdkl |
| Temporal Phenotyping using Deep Predicting Clustering of Disease Progression [Link] | ICML 2020 | alg/ac_tpc |
| Time Series Deconfounder: Estimating Treatment Effects over Time in the Presence of Hidden Confounders [Link] | ICML 2020 | alg/time_series_deconfounder |
| Discriminative Jackknife: Quantifying Uncertainty in Deep Learning via Higher-Order Influence Functions [Link] | ICML 2020 | alg/discriminative-jackknife |
| Frequentist Uncertainty in Recurrent Neural Networks via Blockwise Influence Functions [Link] | ICML 2020 | alg/rnn-blockwise-jackknife |
| Unlabelled Data Improves Bayesian Uncertainty Calibration under Covariate Shift [Link] | ICML 2020 | alg/transductive_dropout |
| Anonymization Through Data Synthesis Using Generative Adversarial Networks (ADS-GAN) [Link] | IEEE | alg/adsgan |
| When and How to Lift the Lockdown? Global COVID-19 Scenario Analysis and Policy Assessment using Compartmental Gaussian Processes [Link] | NeurIPS 2020 | alg/compartmental_gp |
| Strictly Batch Imitation Learning by Energy-based Distribution Matching [Link] | NeurIPS 2020 | alg/edm |
| Gradient Regularized V-Learning for Dynamic Treatment Regimes [Link] | NeurIPS 2020 | alg/grv |
| CASTLE: Regularization via Auxiliary Causal Graph Discovery [Link] | NeurIPS 2020 | alg/castle |
| OrganITE: Optimal transplant donor organ offering using an individual treatment effect [Link] | NeurIPS 2020 | alg/organite |
| Robust Recursive Partitioning for Heterogeneous Treatment Effects with Uncertainty Quantification [Link] | NeurIPS 2020 | alg/r2p-hte |
| Estimating the Effects of Continuous-valued Interventions using Generative Adversarial Networks [Link] | NeurIPS 2020 | alg/scigan |
| Learning outside the Black-Box: The pursuit of interpretable models [Link] | NeurIPS 2020 | alg/Symbolic-Pursuit |
| VIME: Extending the Success of Self- and Semi-supervised Learning to Tabular Domain [Link] | NeurIPS 2020 | alg/vime |
| Scalable Bayesian Inverse Reinforcement Learning [Link] | ICLR 2021 | alg/scalable-birl |
| Nonparametric Estimation of Heterogeneous Treatment Effects: From Theory to Learning Algorithms [Link] | AISTATS 2021 | alg/CATENets |
| Learning Matching Representations for Individualized Organ Transplantation Allocation [Link] | AISTATS 2021 | alg/MatchingRep |
| Explaining by Imitating: Understanding Decisions by Interpretable Policy Learning [Link] | ICLR 2021 | alg/interpole |
| Inverse Decision Modeling: Learning Interpretable Representations of Behavior [Link] | ICML 2021 | alg/ibrc |
| Policy Analysis using Synthetic Controls in Continuous-Time [Link] | ICML 2021 | alg/Synthetic-Controls-in-Continuous-Time |
| Learning Queueing Policies for Organ Transplantation Allocation using Interpretable Counterfactual Survival Analysis [Link] | ICML 2021 | alg/organsync |
| Explaining Time Series Predictions with Dynamic Masks [Link] | ICML 2021 | alg/Dynamask |
| Generative Time-series Modeling with Fourier Flows [Link] | ICLR 2021 | alg/Fourier-flows |
| On Inductive Biases for Heterogeneous Treatment Effect Estimation [Link] | NeurIPS 2021 | alg/CATENets |
| Really Doing Great at Estimating CATE? A Critical Look at ML Benchmarking Practices in Treatment Effect Estimation [Link] | NeurIPS 2021 | alg/CATENets |
| The Medkit-Learn(ing) Environment: Medical Decision Modelling through Simulation [Link] | NeurIPS 2021 | alg/medkit-learn |
| MIRACLE: Causally-Aware Imputation via Learning Missing Data Mechanisms [Link] | NeurIPS 2021 | alg/MIRACLE |
| DECAF: Generating Fair Synthetic Data Using Causally-Aware Generative Networks [Link] | NeurIPS 2021 | alg/DECAF |
| Explaining Latent Representations with a Corpus of Examples [Link] | NeurIPS 2021 | alg/Simplex |
| Closing the loop in medical decision support by understanding clinical decision-making: A case study on organ transplantation [Link] | NeurIPS 2021 | alg/iTransplant |
| Integrating Expert ODEs into Neural ODEs: Pharmacology and Disease Progression [Link] | NeurIPS 2021 | alg/Hybrid-ODE-NeurIPS-2021 |
| SyncTwin: Treatment Effect Estimation with Longitudinal Outcomes [Link] | NeurIPS 2021 | alg/SyncTwin-NeurIPS-2021 |
| Conformal Time-series Forecasting [Link] | NeurIPS 2021 | alg/conformal-rnn |
| Estimating Multi-cause Treatment Effects via Single-cause Perturbation [Link] | NeurIPS 2021 | alg/Single-Cause-Perturbation-NeurIPS-2021 |
| Invariant Causal Imitation Learning for Generalizable Policies [Link] | NeurIPS 2021 | alg/Invariant-Causal-Imitation-Learning |
| Inferring Lexicographically-Ordered Rewards from Preferences [Link] | AAAI 2022 | alg/lori |
| Inverse Online Learning: Understanding Non-Stationary and Reactionary Policies [Link] | ICLR 2022 | alg/inverse-online |
| D-CODE: Discovering Closed-form ODEs from Observed Trajectories [Link] | ICLR 2022 | alg/D-CODE-ICLR-2022 |
| Neural graphical modelling in continuous-time: consistency guarantees and algorithms [Link] | ICLR 2022 | alg/Graphical-modelling-continuous-time |
| Label-Free Explainability for Unsupervised Models [Link] | ICML 2022 | alg/Label-Free-XAI |
| Inverse Contextual Bandits: Learning How Behavior Evolves over Time [Link] | ICML 2022 | alg/invconban |
| Data-SUITE: Data-centric identification of in-distribution incongruous examples [Link] | ICML 2022 | alg/Data-SUITE |
| Continuous-Time Modeling of Counterfactual Outcomes Using Neural Controlled Differential Equations [Link] | ICML 2022 | alg/TE-CDE |
| Concept Activation Regions: A Generalized Framework For Concept-Based Explanations[Link] | NeurIPS 2022 | alg/CARs |
| Benchmarking Heterogeneous Treatment Effect Models through the Lens of Interpretability[Link] | NeurIPS 2022 | alg/ITErpretability |
| Transfer Learning on Heterogeneous Feature Spaces for Treatment Effects Estimation[Link] | NeurIPS 2022 | alg/HTCE-learners |
| Synthetic Model Combination: An Instance-wise Approach to Unsupervised Ensemble Learning[Link] | NeurIPS 2022 | alg/HTCE-learners |
Details of apps and other software is listed below:
| App/Software [Link] | Description | Publication | Code |
|---|---|---|---|
| Adjutorium COVID-19 [Link] | Adjutorium COVID-19: an AI-powered tool that accurately predicts how COVID-19 will impact resource needs (ventilators, ICU beds, etc.) at the individual patient level and the hospital level | - | app/adjutorium-covid19-public |
| Clairvoyance [Link] | Clairvoyance: A Pipeline Toolkit for Medical Time Series | ICML 2021 | clairvoyance repository |
| Clairvoyance2 [Link] | clairvoyance2: a Unified Toolkit for Medical Time Series |
- | clairvoyance2 repository |
| Hide-and-Seek Privacy Challenge [Link] | Hide-and-Seek Privacy Challenge: Synthetic Data Generation vs. Patient Re-identification with Clinical Time-series Data | NeurIPS 2020 competition track | app/hide-and-seek |
Please cite the the applicable papers and van der Schaar Lab repository if you use the software.
See breakdown here.
Copyright 2019-2022 van der Schaar Lab.
This software is released under the 3-Clause BSD license unless mentioned otherwise by the respective algorithms and apps.
See individual algorithm and app directories for installation instructions.
See also doc/install.md for common installation instructions.
See individual algorithm and app directories for tutorials and examples.
Data files (as well as other large files such as saved models etc.) can be downloaded as per instructions in the DATA-*.md (see e.g. DATA-PUBLIC.md) files found in the corresponding directories.
For more information on the van der Schaar Lab’s work, visit our homepage.
See individual algorithm and app directories for references.