These are files used to perform a Signal Tandmobiel analysis 
presented in a paper 

Lesaffre, E., Komarek, A., and Declerck, D. (2005).
An overview of methods for interval-censored data with an emphasis on applications in dentistry.
To appear in: Statistical Methods in Medical Research.
-------------------------------------------------------------------------------------------------

IMPORTANT REMARKS:
==================
If you want to go through these files few changes have to
be carried on (or are recommended) before you run it:
(1) use only part (e.g. n = 100) of Signal Tandmobiel data by adding
    data <- data[1:100, ] if you do not want to wait too long
    for results. To fit one model with a full data set, few hours
    were needed on 2 GHz machine.

    To increase the speed, you might also change definition
    of lambda vector at the beginning of file fittandmob.r. Optimal 
    lambda varied between -2 and -1 for this dataset so there
    is no need to search in other regions.

(2) 'setwd()' commands have to be appropriately modified
    to conform to your working directory.

(3) whenever 'save()' command is used modify appropriately
    its argument file such that it is possible to write
    on disk (or do not save results)

(3) whenever 'load()' command is used you should check 
    that there exists environment to load
    (I used this command to load previously computed results).

FILES:
======
tandmob2.dat
-------------
   data file, see beginning of this file for explanation of variables

fittandmob.r
-------------
   main file to fit different models using 'smoothSurvReg' function
   
plottandmob.r
-------------
   file to create plots shown in the paper

DIRECTORIES:
============
./dataKMM
----------
   this directory contains files with Turnbull's estimates of the survivor
   curves (computed in Splus) for some combinations of covariates. It is used in script
   'plottandmob.r' to add the Turnbull's estimates to plots.      
   