EM_solver {icmstate}R Documentation

EM solver for extended illness-death model (Frydman 1995)

Description

Solves for the cdf and transition intensities using the EM algorithm described in Frydman (1995).

Usage

EM_solver(data_idx, supportMSM, z, lambda, tol = 1e-08)

Arguments

data_idx

List containing data, outputted from msm_frydman

supportMSM

List containing data on the support of the 1->2 transition, output from supportMSM()

z

Initial values for F_{12} and F_{13}, used to initiate the EM alg.

lambda

Initial values for \Lambda_{23}, used to initiate the EM alg.

tol

Tolerance of the EM algorithm. When the change in sum(abs(z)) and sum(abs(lambda)) no longer exceeds tol, the algorithm stops.

Value

beta: Indicator whether Q subset A mu: Value used in the EM algorithm, see Frydman (1995) and Notes. I_Q_in_Lm_tn_star: Indicator whether Q is in [L_m, t_n^*] gamma: Value used in the EM algorithm, see Frydman (1995) and Notes. alpha: Indicator whether Q subset [s_j, Infinity) mu_overline: Value used in the EM algorithm, see Frydman (1995) and Notes. lambda: Intensity for the 2->3 transition z: Mass assigned to the 1->2 and 1->3 transitions iter: Number of iterations required for convergence

References

Frydman, H. (1995). Nonparametric Estimation of a Markov 'Illness-Death' Process from Interval- Censored Observations, with Application to Diabetes Survival Data. Biometrika, 82(4), 773-789. doi:10.2307/2337344


[Package icmstate version 0.1.1 Index]