This project aims to discuss predictive models and interpretability methods in the health field as result of a real application in the obstetric medicine area. Thus, it's expected to predict whether a pregnant diagnosed with gestational diabetes will use insulin previous at one moment to birth and to explain the reason why such outcome occured. Therefore, the first step was evaluate the performance of each model in terms of prediction, and the second, to interpret the one which accuracy was greatest, in this case, the XGBoost model.
Funding source: FAPES.
Cannot be published.
Fitted predictive models - script/fit_models/
Interpreted XGBoost model - script/interpret_models/
result/
util/
shiny/
R, 4.1.0 version, under IDE RStudio