Shaked-g/FinalProject

This project leverages machine learning to analyze Enteral Feeding Intolerance (EFI) using data from ICU patients. The objective is to enhance understanding of EFI and support data-driven treatment decisions by identifying key features and patient subgroups. The analysis includes both supervised and unsupervised learning techniques.

Machine Learning Analysis of Enteral Feeding Intolerance

-Shaked Gofin

Additional Resources:

Poster

Literature research

:fa-file-text-o: :fa-file-pdf-o: :fa-file-pdf-o:

Table of Contents

• 1  Data Overview
• 2  Only Data from the first 3 days
  • 2.1  Data Cleaning
• 3  Machine Learning
  • 3.1  Train Test Split
    • 3.1.1  Stratifying
  • 3.2  Logistic Regression
    • 3.2.1  Evaluation
    • 3.2.2  AUC Curve
    • 3.2.3  Confidence Interval
  • 3.3  Random Forest Classifier
    • 3.3.1  Evaluation
    • 3.3.2  AUC Curve
    • 3.3.3  Feature Importance
    • 3.3.4  Generating the Model on Selected Features
      • 3.3.4.1  Evaluation
      • 3.3.4.2  AUC Curve
  • 3.4  XGBoost - Extreme Gradient Boosting
    • 3.4.1  Evaluate predictions
    • 3.4.2  AUC Curve
    • 3.4.3  Grid Search Cross-Validation
      • 3.4.3.1  Evaluation
      • 3.4.3.2  Feature Importance
      • 3.4.3.3  AUC Curve
  • 3.5  Joint AUC ROC Curve
  • 3.6  Decision Tree
• 4  Unsupervised Learning