Stage-based-Network-Diffusion

Stage-based Network Diffusion (StaND) for Alzheimer's tau progression modeling.

Overview

StaND (Stage-based Network Diffusion) is a computational framework for predicting individual tau progression trajectories in Alzheimer's disease using single-timepoint multimodal neuroimaging data.

Key Features

• Personalized tau progression prediction from single clinic visit
• Hybrid modeling approach combines statistical and biophysical modeling techniques
• Inference of disease origins and kinetic parameters for each individual
• Cross-sectional to longitudinal prediction capability

Prerequisites

• Python 3.8+

Setup

git clone https://github.com/Raj-Lab-UCSF/tau-progression-stand.git
cd tau-progression-stand
pip install -r requirements.txt
Usage
Quick Start
pythonfrom src.stand_framework import StaND


## Repository Structure
tau-progression-stand/
├── data    # Preprocessed ADNI3 data
├── model outputs   # The folder where you will save results
├── src/
    ├── notebooks   # Run through these notebooks sequentially
│       ├── module 1: staging                   # Staging subjects
│       ├── module 2 and 3: biophysical inference and prediction     # Inference of tau seed and parameters and prediction of future spread
    ├── utils/  # Contains all functions required to run notebooks
│       ├── functions_m2.py      
        ├── plotting_functions_m1.py
    ├── models  # Contains the network diffusion model, which is called in the notebooks
├── requirements.txt    
├── LICENSE
└── README.md


## Instructions

Run through each notebook sequentially. The notebooks are set up to call all needed functions and to save results. Notebooks are commented to walk you through the StaND framework. We decided to keep our code in notebook format to enable easy exploration and development.


## Data Access
This repository contains all data you will need to replicate our results, however the framework can be applied to any data set for neurodegenerative diseases. If you wish to directly assess ADNI data,

Register at adni.loni.usc.edu
Request access to ADNI3 dataset
Download required modalities: AV1451 PET, T1 MRI, DTI, cognitive assessments
Follow preprocessing instructions in docs/data_preprocessing.md


## Contributing
We welcome contributions! 


## License
This project is licensed under the Apache License 2.0


## Contact
Primary Contact: Robin Sandell - robin.sandell@ucsf.edu
Lab Website: Raj Lab UCSF


## Acknowledgments
Alzheimer's Disease Neuroimaging Initiative (ADNI) for providing data.


Note: This software is for research purposes only and has not been approved for clinical use.