Home

EasyCIE(GUI) is a rule-based clinical information extraction tool designed for non-NLP(natural language processing) expert users. It a GUI wrapper on top of EasyCIE, an UIMA-based command line version that allows executing on servers.

Downloads

The most recent compiled releases can be found at: https://drive.google.com/drive/folders/0B0hTn1B4kXcPM2hOZDZwY2NOTG8?resourcekey=0-HCxPm8qS6ohgMkf516zhCw&usp=sharing

Although this GUI component is designed to be easily extensible for adapting different databases and wrapping other pipelines without hard coding, this wiki is mainly written as a user guide for EasyCIE(GUI).

EasyCIE(GUI) functionalities

• Import: Import documents from files into database. Import eHOST or Brat annotations as the reference standard. Import ontologies (https://github.com/Blulab-Utah/resource_ontologies)
• RunEasyCIE: Execute the EasyCIE pipeline on imported data
• Compare: Compare the pipeline output between different version or against the imported reference standard.
• Debug: Use a snippet of text to tackle the errors
• Export: Export the EasyCIE output into eHOST, Brat or UIMA XMIformat.

EasyCIE NLP components

EasyCIE is under active development. The components are expanding rapidly. In the current released version, the built-in NLP pipeline includes the following components:

• RuSH: an efficient, reliable, and easy adaptable rule-based sentence segmentation solution.
• FastNER: a speed-optimized dictionary-based name entity recognition solution.
• FastContext: an optimized Java implementation of the ConText algorithm (https://www.ncbi.nlm.nih.gov/pubmed/23920642).
• FeatureInferenceAnnotator: allows customizable conversion among UIMA types.
• DocInferenceAnnotator: allows to draw document level conclusion from mention level annotations using customizable rules.

Configure your pipeline

All the components below are configurable through an excel file:

• Configure FastNER
• Configure FastContext
• Configure FeatureInferenceAnnotator
• Configure DocInferenceAnnotator
• Configure RuSH

Suggestions & Comments

If you have any issues, suggestions or comments regarding using EasyCIE, feel free to leave a messages in issues.

Acknowledgement

Special thanks to Danielle Mowery for helping the EasyCIE tutorial at ICHI2017. The tutorial material can be accessed from here.

All the demonstration examples are pulled from MIMIC II demo dataset. Thanks to PhysioNet for making this de-identified dataset publically available for free.