The repo contains a visualizer that lets clinicians or researchers inspect individual admissions from the MIMIC-IV dataset, browse relevant notes and codes, and interactively plot patient data over time.
- Connects to a local MIMIC-IV MySQL database and a MongoDB instance for discharge notes and ecg metadata.
- Provides a web interface where you enter a subject_id and pick a hospital admission to explore. Patient demographics, insurance, language, and admission/ICU details are displayed.
- Shows ICD diagnoses, procedures, and any discharge notes retrieved from MongoDB. ICU stays are visualized on an interactive timeline, with a marker for patient death if applicable.
- Lists all available data items across chart events, lab results, and prescriptions. You can filter or sort items, add them to a selection, and scroll through paginated results.
- Allows you to pick a time range during the admission and plot the selected items as trends, events, or timelines, enabling quick inspection of vital signs, lab values, or medication orders.
- New Include item
ECGfrom MIMIV-IV-ECG and visualization of ECG waveforms.
- Following the instruction in mimic-code to populate the mimic-iv dataset to mysql databse.
- Download the MIMIC-IV-ECG and unzip to
/path/to/mimic-ecg. - Run
pip install -r requirement.txtto install the dependency - Run script
scripts/load_mimic_note_to_mongo.pyto populate the mimic-iv-node discharge notes to mongo - Run script
scripts/load_mimic_ecg_to_mongo.pyto populate the mimic-iv-ecg ECG machine_measurement to mongo - Run
streamlit run app.py -- --mysql-host YOUR_HOST --mysql-user YOUR_USER --mysql-password YOUR_PASSWORD --mongo-uri YOUR_MONGO_URI --ecg-base-folder /path/to/mimic-ecgto start the server. The ECG base folder should point to the directory that contains thefiles/pNNNN/...tree. - Open
http://localhost:8501/?path=ecg/pXXXXXXXX/sZZZZZZZZ(replace with the subject and study identifiers) to view an ECG waveform. If you navigate to?path=ecgyou can use the on-page input to enter a locator manually.
