Mixed Multivariate Cumulative Incidence Functions
Computes the Log Cholesky Decomposition and the Inverse
Sets up an Object to Compute the Log Composite Likelihood
Fits a Mixed Competing Risk Model
Evaluates the Log Composite Likelihood and its Gradient
Computes Marginal Measures Using Two Observations
Computes Marginal Measures for One Observation
Computes the Sandwich Estimator
Finds Staring Values
Fits the mixed cumulative incidence functions model suggested by <doi:10.1093/biostatistics/kxx072> which decomposes within cluster dependence of risk and timing. The estimation method supports computation in parallel using a shared memory C++ implementation. A sandwich estimator of the covariance matrix is available. Natural cubic splines are used to provide a flexible model for the cumulative incidence functions.