Skip to content

Latest commit

 

History

History
executable file
·
44 lines (33 loc) · 3.42 KB

README.md

File metadata and controls

executable file
·
44 lines (33 loc) · 3.42 KB
Raw

MIMIC-IV Data Extraction

Here we modified codes from the mimic3-benchmarks and MedFuse for extract and process EHR data for our task.

Download the MIMIC-IV

We do not provide the MIMIC-IV data itself. You must acquire the data yourself from https://physionet.org/content/mimiciv/2.0/. Specially, download and extract the CSV files.

Pre-process the EHR data

  1. The following command takes MIMIC-IV CSVs, generates one directory per SUBJECT_ID and writes ICU stay information to data/{SUBJECT_ID}/stays.csv, diagnoses to data/{SUBJECT_ID}/diagnoses.csv, and events to data/{SUBJECT_ID}/events.csv.
    ```bash cd ./mimic4extract

    mimic_iv_csv_dir="/path/to/mimic_iv_csv" # e.g. /root/autodl-tmp/0807_version/physionet.org/files/mimiciv/2.0/
mimic_iv_subjects_dir="path/to/save/mimic-iv/subjects/ehr" # e.g. /root/autodl-tmp/dynamic-prediction/data/mimic_iv_subjects_dir

python -m mimic3benchmark.scripts.extract_subjects_iv $mimic_iv_csv_dir $mimic_iv_subjects_dir
```

Meanwhile, after this process ICU stays of all the subjects has been save in $mimic_iv_subjects_dir/all_stays.csv. And phenotype labels of all the subjects has been save in $mimic_iv_subjects_dir/phenotype_labels.csv.

  1. The following command attempts to fix some issues (ICU stay ID is missing) and removes the events that have missing information. About 80% of events remain after removing all suspicious rows (more information can be found in mimic3benchmark/scripts/more_on_validating_events.md).

    ```bash
python -m mimic3benchmark.scripts.validate_events $mimic_iv_subjects_dir
```

  2. The next command breaks up per-subject data into separate episodes (pertaining to ICU stays). Time series of events are stored in {SUBJECT_ID}/episode{ICU_STAY_ID}_timeseries.csv while episode-level information (patient age, gender, ethnicity, height, weight) and outcomes (mortality, diagnoses) are stores in {SUBJECT_ID}/episode{ICU_STAY_ID}.csv. bash python -m mimic3benchmark.scripts.extract_episodes_from_subjects $mimic_iv_subjects_dir

  3. After the steps above, each subject in $mimic_iv_subjects_dir has the directory tree as follows:

{SUBJECT_ID}
|---diagnoses.csv  #(Fields: subject_id,hadm_id,seq_num,icd_code,icd_version,long_title,stay_id,HCUP_CCS_2015,USE_IN_BENCHMARK)
|---events.csv  #(Fields:subject_id,hadm_id,stay_id,charttime,itemid,value,valuenum)
|---stays.csv #(Fields:subject_id,hadm_id,stay_id,last_careunit,intime,outtime,los,admittime,dischtime,deathtime,gender,anchor_age,dod,age,mortality_inunit,mortality,mortality_inhospital)
|---episode{ICU_STAY_ID_1}.csv       #(Fields:Icustay,Gender,Age,Height,Weight,Length of Stay,Mortality,Diagnosis I169,Diagnosis I509,Diagnosis I2510,Diagnosis I4891...)
|---episode{ICU_STAY_ID_1}_timeseries.csv #(Fields:Hours,Capillary refill rate,Diastolic blood pressure,Fraction inspired oxygen,Glascow coma scale eye opening,Glascow coma scale motor response,Glascow coma scale total,Glascow coma scale verbal response,Glucose,Heart Rate,Height,Mean blood pressure,Oxygen saturation,Respiratory rate,Systolic blood pressure,Temperature,Weight,ph)
|---episode{ICU_STAY_ID_2}_timeseries.csv
|---episode{ICU_STAY_ID_2}.csv  # A subject might have multuples icu stays.