Inferring mood disorder symptoms from multivariate time-series sensory data

Code for workshop paper "Inferring mood disorder symptoms from multivariate time-series sensory data" at the NeurIPS 2022 Workshop on Learning from Time Series for Health.

@article{
    li2022inferring,
    title={Inferring mood disorder symptoms from multivariate time-series sensory data},
    author={Bryan M. Li and Filippo Corponi and Gerard Anmella and Ariadna Mas and Miriam Sanabra and Diego Hidalgo-Mazzei and Antonio Vergari},
    journal={NeurIPS 2022 Workshop on Learning from Time Series for Health},
    year={2022},
    url={https://openreview.net/forum?id=awjU8fCDZjS}
}

Installation

• Create a new conda environment with Python 3.8.

  conda create -n timebase python=3.8

• Activate timebase virtual environment

  conda activate timebase

• Install all dependencies and packages with setup.sh script, works on both Linus and macOS.

  sh setup.sh

Dataset

• See dataset/README.md regarding data availability and the structure of the dataset.

Train regression model

• To train a BiLSTM regression model with GRU embeddings

  python regression_train.py --output_dir runs/001_test_run --regression_mode 1 --qc_mode 1 --time_alignment 0 --embedding_type 0 --model bilstm --epochs 200 --verbose 1
  

• Use python regression_train.py --help to see all options.
• TensorBoard visualization

  tensorboard --logdir runs/001_test_run