Case Study on Survival Analysis: Prediction of Cardiovascular Events

Abstract

Survival is an important long-term outcome in prognostic research, including medical areas such as cardiovascular disease and oncology. We consider a model for the occurrence of vascular events in patients with symptomatic cardiovascular disease. Patient data were from the Second Manifestations of ARTerial disease (SMART) study. We go through the seven steps of the checklist for developing valid prediction models, as presented in Part II. Specific focus is on the combination of models specification and estimation with the LASSO in combination with multiple imputation of missing values. The data set and R code are made available at the book’s website (www.clinicalpredictionmodels.org).