WARNING: there is no any guarantee on the use of this program.
**************************
Usage:

Step 1. Create object file myicm.o from C program myicm.c:
  gcc -c myicm.c

Step 2. Start Splus and in Splus
> dyn.load("myicm.o") 
> source("CVbandw.s")  #functions for bandwidth selection
> source("kernel2test.s") #using kernel smoother; result stored in a 
                        #subdirectory named KERNEL, under the current dir
> kernel2test("mydata", lbandw=0.1, ubandw=1.2, nfold=10)
> source("logspline.s") #using log-spline; result stored in a subdirectory
                        # named LGSPLN1, under the current dir
> test2Logspline("mydata")
> source("NPMLE2test.s") #using NPMLE; result stored in a subdirectory
                        # named NPMLE, under the current dir
> NPMLE2test("mydata")

**************************
Input data format (e.g. in "mydata" file):
   An input datafile contains multiple lines: each line is an 
observation in the form of
   Li  Ri  j
where the event is censored between Li and Ri (Li=0 for left-censoring;
Ri=99999 for right-censoring; Ri=Li+epsilon (i.e. a tiny number) for an
exact observation), and j=0 or 1 represents one of the two samples being
compared.

*************************
Reference:
  Pan, W. (2000) ``Smooth Estimation of the Survival for Interval Censored 
    Data". Statistics in Medicine, 19, 2611-2624
*************************
Wei Pan

weip@biostat.umn.edu
http://www.biostat.umn.edu/~weip

Division of Biostatistics, University of Minnesota,
A460 Mayo, MMC 303, Minneapolis, MN 55455

April 2001.