smoothROCtime: an R package for time-dependent ROC curve estimation
| Autor: | Sonia Pérez-Fernández, Susana Díaz-Coto, Pablo Martínez-Camblor |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Statistics and Probability
Receiver operating characteristic business.industry Computer science 05 social sciences Kernel density estimation Estimator Probability density function Bivariate analysis 01 natural sciences 010104 statistics & probability Computational Mathematics Software 0502 economics and business Sensitivity (control systems) 0101 mathematics Statistics Probability and Uncertainty business Algorithm 050205 econometrics Variable (mathematics) |
| Zdroj: | Computational Statistics. 35:1231-1251 |
| ISSN: | 1613-9658 0943-4062 |
| DOI: | 10.1007/s00180-020-00955-7 |
| Popis: | The receiver operating characteristic (ROC) curve has become one of the most used tools for analyzing the diagnostic capacity of continuous biomarkers. When the studied outcome is a time-dependent variable two main generalizations have been proposed, based on properly extensions of the sensitivity and the specificity. Different procedures have been suggested for their estimation mainly under the presence of right censorship. Most of them have been implemented, as well, in diverse types of software, including R packages. This work focuses on the R implementation for the smooth estimation of time-dependent ROC curves. The theoretical connection between them through the joint distribution function of the biomarker and time-to-event variables prompts an approximation method: considered estimators are based on the bivariate kernel density estimator for the joint density function of the bidimensional variable (Marker, Time-to-event). The use of the package is illustrated with two real-world examples. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink