How saharshps/alzheimer-s-disease-classification- works
This is a Streamlit application for predicting Alzheimer’s disease from patient-reported demographic, health, and cognitive features. The main workflow is: a user enters values, the app displays them, and—on request—calls a serialized LightGBM model and displays a prediction. The README also describes an offline analysis and modeling workflow covering EDA, preprocessing, model comparison, evaluation, cross-validation, tuning, and ROC-AUC analysis. Those capabilities are documented, but notebook internals were not sampled, so their detailed data flow is not asserted.
Main entry points: app.py Prediction interface · alzheimers_disease_data.csv Patient dataset · model.ipynb Data preprocessing · alzheimer_model.joblib Trained model
Components 11
Prediction app
- Prediction interface
app.py - Trained model
alzheimer_model.joblib - Prediction result
app.py
Data analysis
- Patient dataset
alzheimers_disease_data.csv - Exploratory analysis
visualization.ipynb
Model development
- Data preprocessing
model.ipynb - Model comparison
model.ipynb - Model evaluation
model.ipynb - Validation and tuning
model.ipynb - LightGBM selection
model.ipynb
Other components
- User
Connections 10
- User → Prediction interface: enters features
app.py - Prediction interface → Trained model: loads and predicts
app.py - Prediction interface → Prediction result: displays prediction
app.py - Patient dataset → Exploratory analysis: analyzes data
- Patient dataset → Data preprocessing: prepares data
- Data preprocessing → Model comparison: feeds models
- Model comparison → Model evaluation: compares metrics
- Model evaluation → Validation and tuning: informs selection
- Validation and tuning → LightGBM selection: supports selection
- LightGBM selection → Trained model: exports model