This is a project in association with the Global Women in Data Science Conference (March 2, 2020). WIDS is hosting this datathon challenge in which we will use patient health data provided by MIT GOSSIS (Global Open Source Severity of Illness Score) Initiative to predict patient survival. A more in-depth description can be found here: https://www.kaggle.com/c/widsdatathon2020/overview
Our data includes:
- Over 130,000 hospital (ICU) visits from patients over a one-year timeframe
- Data is collected from > 200 hospitals around the world
- Labeled traininng data is provided
The data is provided as 5 csvs:
- training_v2.csv - labeled training data (91,713 encounters in the training set)
- unlabeled.csv - test data set used for competition scoring
- samplesubmission.csv - a sample submission file in the correct format
- solution_template.csv - a list of all the rows (and encounters) that should be in your submissions
- WiDS Datathon 2020 Dictionary.csv - supplemental information about the data
Specifically, our data includes:
- Identifiers
- Demographic information
- Covariate information
- Laboratory results
- Comorbidity
- Diagnosis upon admission
The Data can be downloaded here: https://www.kaggle.com/c/widsdatathon2020/data