In this project, we aim to reproduce the results of the paper "Improving Medical Predictions by Irregular Multimodal Electronic Health Records Modeling" by Xinlu Zhang, Shiyang Li, Zhiyu Chen, Xifeng Yan, and Linda Petzold.
Paper: https://arxiv.org/abs/2210.12156
Repo: https://github.com/XZhang97666/MultimodalMIMIC
For this project we have cloned the original repo, MultimodalMIMIC, as well as ClinicalNotesICU and mimic3-benchmarks into this repo as modules to use on out Notebook ProjectNotebook.ipynb.
- Clone this repo
- Install Python 3 if not already installed
- Install Jupyter if not already installed
- Create a virtual environment, we recommend using conda
- Install the required packages using the Requirements.txt file using the following command:
pip install -r Requirements.txt- Open GlobalConfig.py and set the PROJECT_BASE_PATH to point to the root of this project
- Open ProjectNotebook.ipynb, read and edit the variables for paths and credentials as needed.
These are some of the main configs to set:
- Set the flags on the Config tasks to execute section to enable the downloading, decompressing, and processing tasks.
- Set credentials to physionet on the Download dataset section. if intend to download the dataset from the notebook.
- Set the paths to the dataset folders on the Configuring imports and directories section (only needed if the dataset is not in the default paths from the notebook).
.
├── ClinicalNotesICU
│ ├── models
│ └── scripts
├── mimic3-benchmarks
│ ├── data
│ │ ├── decompensation
│ │ ├── in-hospital-mortality
│ │ ├── length-of-stay
│ │ ├── multitask
│ │ ├── phenotyping
│ │ └── root
│ │ ├── test_text_fixed
│ │ └── text_fixed
│ ├── mimic3benchmark
│ │ ├── evaluation
│ │ ├── resources
│ │ ├── scripts
│ │ └── tests
│ │ └── resources
│ └── mimic3models
│ ├── decompensation
│ │ └── logistic
│ ├── in_hospital_mortality
│ │ └── logistic
│ ├── keras_models
│ ├── length_of_stay
│ │ └── logistic
│ ├── multitask
│ ├── phenotyping
│ │ └── logistic
│ └── resources
└── MultimodalMIMIC
├── Data
│ ├── ihm
│ └── irregular
└── run
└── TS_Text