Plots
POP-Pe is a a machine learning tool for Prediction of sustained OPioid use in Pediatrics, is was build based on data from the study Opioid trends and risk factors for sustained use among children and adolescents in Israel: a retrospective cohort study. Full details regarding the cohort and the data can be found here.
The full description of the model fitting procces can be found in our paper Prediction of sustained opioid use in children and adolescents using machine learning
Three models were train and tested for different follow-up periods: one, two and three years. The one year models had the best discrimination and calibration and thus used for the final model. Code can be found on my GitHub
This calculator predict the risk of children and adolescence, age 19 and under, who are naive to opioids to devolp a sustained use pattern. All demographic characteristics and medical data referese to the preceding year to the first opioid purchase.
The model was trained using xgboost and tidymodels in R, and than calibrated using Shape Constrained Additive Models scam.
NOTICE: