Fits mixture cure rate models by the Expectation-Maximization (EM)
algorithm. The incidence component (the probability of being uncured)
accepts the logit, probit, cauchit, power logit and reversed power logit
link functions, and the
latency component accepts the exponential, Rayleigh, Weibull, log-normal,
log-logistic and inverse Gaussian distributions. The package provides
parameter estimates with standard errors, simulation of data from the
model, and diagnostic tools based on residuals and simulated envelopes.
The methods build on Berkson and Gage (1952)