02-19-2017 03:26 PM
Please experts, I am trying to generate survival times for recurrent events. The event times follows the Cox's proportional hazard model in which a patient can have more than one event of the same time over time. The are two covariate in the model. one of the covariate a class variable (0,1) representing treatment group indictor. The other covariate is a count variable; that is 0, 1,2, 3,4.... very large counts are not usually possible, and this variable contain about 50 percent zeros with other values been 1, 2, 3, 4,5. The count covariate is susposed to represent multiple event times before the study, so I am summing up the event time to obtain the count for simplicity.
The model looks like this:
T=lambda0*exp(beta1 +beta2). Beta1 is the treatment indicator; beta2 is an explanatory count variable the is zero inflated Poison. T is the recurrent survival times. Its unusual to have more than 7 events in the time period under consideration.
Any help will be appreciated.