ChangchangXu-LTRI/Mixcure: Maximum (penalized) likelihood estimation of Weibull mixture cure model

Maximum likelihood estimation of the usual likelihood or Firth-type penalized likelihood of a mixture cure model. The mixture cure model was constructed to jointly model time-to-event (latency) and probability of being event-free (incidence).The latency part of the model is a conventional survival regression with the hazardous baseline approximated by a Weibull function, and the incidence part of the model utilizes a logistic regression. Parameter optimization of both log-likelihoods are via a Newton-type algorithm for non-linear minimization. As the means for hypothesis testing and interval estimation, two alternatives were included: i) Wald-type statistics and Wald-type confidence intervals; ii) likelihood ratio statistics and profile likelihood confidence intervals.

install.packages("remotes")
remotes::install_github("ChangchangXu-LTRI/Mixcure")