R packages for prospective biomarker validation with time-to-event data



Biomarker evaluation with survival data

  • survAccuracyMeasures: Estimate the AUC, TPR(c), FPR(c), PPV(c), and NPV(c) for for a specific timepoint and marker cutoff value c using non-parametric and semi-parametric methods. Bootstrap standard errors and confidence intervals are also computed.

  • survCompetingRisk: Evaluate the prognostic accuracy of a marker with multiple competing risk events. Functions to calculate the AUC, ROC, PPV, and NPV are provided. A discrete covariate Z, if available, can be included.

  • survNRI: Calculate the net reclassification index (NRI) statistic for survival data using five different estimators.


Prediction with longitudinal biomarkers

  • partlyconditional Provides functions to fit partly conditional models, a class of models useful for clinical contexts where long term follow-up information is available on a patient population, along with multiple measures of patient health and other biological markers collected across time. Interest lies in modeling a patients’ future risk of an adverse event, given survival to time s, as a function of longitudinal marker history recorded up to time s.

  • longsurvAccuracyMeasures: Evaluate a partly conditional model fit using the R package  'partlyconditional'. Given validation data consisting of a time-to-event outcome and longitudinal marker information on a set of individuals up to a fixed time window, this package includes functions to assess model calibration and prognostic accuracy at a specified future landmark prediction time. Point estimates, along with bootstrap confidence intervals, are provided for the following accuracy measures: AUC, ROC (TPF/FPF), PPV, NPV, PCF (proportion cases followed), and PNF (proportion needed to follow).


  • longsurvMarkerTwoPhase:  Functions to evaluate the prognostic accuracy of a partly conditional model fit using a longitudinal marker from a case-cohort (CCH) or a nested case-control (NCC) sample, using methods described in "Evaluating longitudinal markers under two-phase study designs’" by Marlena Maziarz, Tianxi Ci, Li Qi, Anna Lok and Yingye Zheng.



Two-phase biomarker study design

  • survMarkerTwoPhase: Evaluate the prognostic accuracy of a marker under two phase designs (case-cohort and nested case control designs).

  • AIPWmeasures: Estimate measures of predictive accuracy using augmented inverse probability weights for two-phase biomarker validation studies.


Power Calculations

Shiny web applications to simulate power for prognostic biomarker validation studies with time to event data.