Skip to content

Latest commit

 

History

History
46 lines (39 loc) · 1.55 KB

README.md

File metadata and controls

46 lines (39 loc) · 1.55 KB

MOTIMGluPred

This is the repository for our work glucose excursion prediction using meta-optimised time-index model.

Prepare Data

Raw data of the two datasets should be acquired from their raw repositories.

  1. prepare the following structure.
storage/
+-- datasets/
|   +-- ohiot1dm/    
|   |   +-- OhioT1DM-training/
|   |   +-- OhioT1DM-testing/
|   |   +-- OhioT1DM-2-training/
|   |   +-- OhioT1DM-2-testing/
|   +-- umt1dm/
|   |   +-- unprocessed_cgm_data.xlsx
  1. Preprocessing the data using the following commands, which would output the preprocessed data files {ID}_{train/test}.csv under storage/datasets/{dataset}/preprocessed/.
python -m storage.datasets.ohiot1dm.preprocess
python -m storage.datasets.umt1dm.preprocess

Requirements

Dependencies for this project can be installed by

pip install -r requirements.txt

Usage

One can directly run our experiment using the script train_model.sh, where the GPU device number and the experiment name (config file name in experiments/configs) should be provided as arguments. For example:

bash train_model.sh 0 ohiot1dm_self

The results will be saved to storage/experiments/.

Acknowledgements

Our implementation is based on resources from the following repository, we thank the original authors for open-sourcing their work.