Machine Learning for Depression Classification

Implementing ML models to classify depression levels based on questionnaire data.

Overview

Developed a machine learning pipeline for classifying depression levels using Burns Depression Checklist (BDC) scores.

Links:

Features

  • Preprocessing: Used SelectKBest, mRMR, and Boruta for feature selection.
  • Models: Implemented KNN, XGBoost, Gradient Boosting, AdaBoost, and Random Forest classifiers.
  • Evaluation: Achieved high accuracy and F1 scores with detailed confusion matrix analysis.

Current Status

  • Extending the model for real-world deployment in healthcare applications.
  • Exploring interpretability techniques like SHAP to understand model predictions.