This repository presents the implementation of different machine learning architectures to determine the efficacy of the Acute Physiology andChronic Health Evaluation (APACHE) IV score as well as the patient characteristics that comprise it to predict the discharge destination for critically ill patients within 24 hours of ICU admission.

---

Overall Architecture

---

This repository contains the following components:

• Handling Missing Data

  • Missingness indicator
  • MICE Imputation

• Handling Class Imbalance

  • Class_weight
  • SMOTE Oversampling
  • SMOTEENN
  • SMOTE-NC
  • No-adjustment

• Predictive Models

  • Logistic Regression
  • Random Forest
  • Support Vector Machines
  • K-Nearest Neighbours
  • XGBoost
  • AdaBoost
  • ExtraTrees
  • Deep Neural Networks
  • Hierarchical Classifiers
  • Staked Ensemble Models
  • Mix Clustering-Classification Models