van der Schaar Lab


Visit our GitHub – it’s the central hub for all our software!

Try out our interactive demonstrators to see how machine learning-powered tools can be transformative in the hands of clinicians!

Real-world Impact

The van der Schaar Lab is committed to fostering a real-world impact through open-source software. With trailblazing tools such as AutoPrognosis 2.0, which harnesses the potential of AutoML for healthcare, and SynthCity, a pioneering solution for generating and evaluating synthetic data for privacy, fairness and data augmentation, we are redefining healthcare paradigms. The lab’s contributions also extend to the realms of data imputation with HyperImpute, time-series exploration via TemporAI and Clairvoyance, enhancing AI transparency with our Interpretability Suite, and driving the data-centric AI agenda with Datagnosis. Our relentless pursuit of excellence and drive for tangible societal change is further underscored by our vast repository of research code available on GitHub, embodying our dedication to knowledge dissemination and open collaboration.

Key Software Projects

AutoPrognosis 2.0 is the van der Schaar Lab’s newest framework offering the healthcare community a tool that utilizes the power of AutoML in a flexible and interpretable way.

HyperImpute simplifies the selection process of a data imputation algorithm for your ML pipelines. It includes various novel algorithms for missing data imputation and is compatible with sklearn.

SynthCity is a powerful library for generating and evaluating synthetic tabular data for privacy, fairness and data augmentation.

Clairvoyance is our ground-breaking pipeline toolkit for medical time-series data that represents the culmination of years of research, development, and real-world testing we announced in May 2020.

TemporAI is a Machine Learning-centric time-series library for medicine. The tasks that are currently of focus in TemporAI are: time-to-event (survival) analysis, treatment effects, and prediction. TemporAI enhances and builds upon Clairvoyance.

Datagnosis is a Data-Centric AI library for measuring hardness categorization. It provides several state-of-the-art methods for data hardness categorization in an easy-to-extend architecture.

The Interpretability Suite is a collection of Machine Learning interpretability methods – the methods aim to provide an insight into why a model has made a given prediction. It also provides a common interface and a sandbox web app.

All Other Software and Research Code

You can find an extensive collection of our existing software on our GitHub org.

Historic Projects