Dynamichazard: Dynamic Hazard Models Using State Space Models

Benjamin Christoffersen

Research output: Contribution to journalJournal articleResearchpeer-review

1 Downloads (Pure)


The dynamichazard package implements state space models that can provide a computationally efficient way to model time-varying parameters in survival analysis. I cover the models and some of the estimation methods implemented in dynamichazard, apply them to a large data set, and perform a simulation study to illustrate the methods' computation time and performance. One of the methods is compared with other models implemented in R which allow for left-truncation, right-censoring, time-varying covariates, and timevarying parameters.
Original languageEnglish
JournalJournal of Statistical Software
Issue number7
Pages (from-to)1-38
Number of pages38
Publication statusPublished - Sep 2021


  • Survival analysis
  • Time-varying parameters
  • Extended Kalman filter
  • EM algorithm
  • Unscented Kalman filter
  • Parallel computing
  • R
  • Rcpp
  • RcppArmadillo

Cite this