Note: If you are referred from ICE-NODE paper, please follow the relevant instructions on the following snapshot of the codebase: MLHC 2022 version.

Roadmap

• Pipeline validators.
• Integrate consort diagramming in the pipeline.
• Work with the pytable library directly instead of the pandas library.
• Implement packed representations for the tvx_ehr to improve the compressibility of the data.
• Implement a scheme manager object to handle schemes and codemaps, instead of using global variables.
• lib.ehr.tvx* test.
• lib.ehr.coding_scheme.CodeMap test.
• lib.ehr.* documentation / document edge cases tested.
• lib.ehr custom exceptions / adapt tests.
• FHIR resources adaptation.
• Support for SNOMED-CT.
• CLI for running pipelines.
• GUI for configuring the dataset and the tvx_ehr.
  • Pipeline 10 + 10 steps.
  • Selection of dataset CodingScheme space.

Coverage

Branch	Coverage
main
dev

Citation

To cite this work, please use the following BibTex entry:

@article{Alaa2022ICENODEIO,
  title={ICE-NODE: Integration of Clinical Embeddings with Neural Ordinary Differential Equations},
  author={Asem Alaa and Erik Mayer and Mauricio Barahona},
  journal={ArXiv},
  year={2022},
  volume={abs/2207.01873}
}