1: \begin{abstract}

2: We consider linear transformation models applied to right censored survival

3: data with a change-point in the regression coefficient based on a covariate

4: threshold.  We establish consistency and weak convergence of the

5: nonparametric maximum likelihood estimators.  The change-point parameter

6: is shown to be $\,n$-consistent, while the remaining parameters are shown to 

7: have the expected root-$n$ consistency. We show that the 

8: procedure is adaptive in the sense that the non-threshold parameters are 

9: estimable with the same precision as if the true threshold value were known.

10: We also develop Monte-Carlo methods of inference for model parameters

11: and score tests for the existence of a change-point.

12: A key difficulty here is that some of the model

13: parameters are not identifiable under the null hypothesis of no change-point.

14: Simulation studies establish the validity of the proposed 

15: score tests for finite sample sizes.

16: \end{abstract}

17: