Identifying Episodes of Hypovigilance in Intensive Care Units Using Routine Physiological Parameters and Artificial Intelligence: a Derivation Study. Open Code and Dataset

The purpose of this project is to detect hypogilance using the EVEILS database.

Database is ICU data from Hôtel-Dieu De Lévis , Québec, Canada. Please cite us if you use either the data or code. 

This code was written during Raphaëlle Giguère Msc in Computer Science. The goal of her project is to detect hypovigilance using machine learning in the ICU. In this repository, you have the data set before preprocessing:

df_hypovigilance : Contains the hours, date and value of the vigilance level, using either the RASS or Ramsay and already converted using the thresholds shown in the paper.

raw_df : Contains the raw values from the gateway for each participant. All of the identifying values have been removed.

At the end of the preprocessing_anonymous script, you should generate a new dataset called "df_final". This dataset is used for the training_model script.

The cross validation employs groups of random size meaning the results might differ from time to time but should stay consistent.