Sanfilippo Machine Learning

This repository contains code and models for the Sanfilippo Paper, focused on distinguishing Healthy vs MPSIIIA and Healthy vs MPSIIIA + stress using high content cell imaging analysis and raw image classification.

Folder Structure

• feature_models/
  Contains XGBoost models trained on features extracted from cell imaging analysis (Harmony software).
  Tasks:

  • Healthy vs MPSIIIA
  • Healthy vs MPSIIIA + stress
    Dependencies:
  • Python 3.11
  • scikit-learn
  • XGBoost
  • pandas
  • matplotlib
  • Hyperopt
  • seaborn
  • SHAP

• Image_models/
  Contains CNN models for classifying raw images.
  Tasks:

  • Healthy vs MPSIIIA
  • Healthy vs MPSIIIA + stress
    Dependencies:
  • Python 3.11
  • PyTorch
  • Optuna
  • matplotlib
  • numpy
  • pandas
  • seaborn

Getting Started

1. Clone the repository.
2. Install dependencies for each folder (see above).
3. Refer to folder-specific README or scripts for training and evaluation instructions.

Citation

If you use this code or models, please cite the Sanfilippo Paper.