Awesome

PopNet

The source code for PopNet: Real-Time Population-Level Disease Prediction with Data Latency

Requirements

• Install python, pytorch. We use Python 3.8, Pytorch 1.1.
• Install dgl.
• If you plan to use GPU computation, install CUDA.

Synthetic dataset

We provide the synthetic dataset in data directory. The synthetic dataset have 1015 locations and the sequence length is 63. We also provide the generated location graph in data/g. Use following codes to load the dataset

import pickle
FILE = pickle.load(open('./data/FILENAME','rb'))

Test PopNet

You provide the testing code in the train-popnet.ipynb notebook. We also provide a trained model in the save directory. You can run the code to get the performance of PopNet on the synthetic dataset.

Train PopNet

We provide the training code for PopNet in the notebook. You can use them to train our model on other dataset.