Toolboxes & Code Repositories

TreeNet

Library to unify temporal series with hierarchical semantic information through dynamic path aggregation on tree-structured data

software
Description:

TreeNet is a graph learning method that uses dynamic path aggregation on tree-structured Electronic Health Records (EHR), enabling deep information propagation and high-velocity temporal integration. TreeNet is built to be robust; it supports both static non-numerical concepts and dynamic time-series data, outperforming traditional LSTMs and standard graph neural networks in predictive healthcare tasks like mortality prediction.

More Information:

  • Originates from the AI-Driven Data Science research challenge (FAIR 2019–2023).

  • Responsible Research Lead: Sofie VanHoecke

  • Authors: David vander Mijnsbrugge, Sandeep Ramachandra, Femke Ongenae, Sofie Van Hoecke 

  • Programming Language: Python

  • Link to the repository

  • Categories: Networks and graphs

Publications:

  • Vander Mijnsbrugge, D. (2024). Context aware deep learning for predictive healthcare. Ghent University. Faculty of Engineering and Architecture, Ghent, Belgium.