Model Personalizatin with Static and Dynamic Patients' Data

Data from MIMIC-Extract

• run ./data/data_extraction_mimic_extract.py to prepare the bp data for experiments. Put the output under 'DATA_DIR/data/bp/' or change the directory in data_utils
• Note that for this step you need to have access to MIMIC-III
• The models for generating the amplitude and phase for the SX dataset can be found under ./data/sx/
• The generic models for each dataset and setting are included under their directory

Experiments

• run w_static.py for the w/ static models, wo_static.py for the w/o static models under './experiments/DATASET_NAME' to get the resulls for ml models and tl models
• run eval_train.py under the same directory to generate the task-specific models for the training set
• Correspondingly, run reconst_gen.py under each directory to construct models based on static attributes using td-maml

Citation

@INPROCEEDINGS{youssef-etal-2022-personalization,
  author={Youssef, Paul and Schlötterer, Jörg and Imangaliyev, Sultan and Seifert, Christin},
  booktitle={2022 IEEE International Conference on Data Mining Workshops (ICDMW)}, 
  title={Model Personalization with Static and Dynamic Patients' Data}, 
  year={2022},
  pages={324-333},
  doi={10.1109/ICDMW58026.2022.00051}}