# RUSSIAN ROULETTE  

# (changeable parameters)

n.bullets=1; n.chambers=6; n.occasions=7;


# theoretically..

prob.survive.n.occasions = (1-n.bullets/n.chambers)^n.occasions; prob.survive.n.occasions


# by simulation.. 100 persons [0=survived, 1=dead]

n.persons = 100;
lifeline=matrix(rep("",n.persons),nrow=n.persons);
outcome=c("L","D");
for (person in 1:n.persons)
{
 alive=1;
 for (occasion in 1:n.occasions) 
  {
   if (alive==1) {
          result=rbinom(1,1,1/n.bullets/n.chambers);
          lifeline[person,1]=paste( lifeline[person,1],outcome[result+1] );
          alive=(result==0)
              }
  }
}
lifeline;

# sorted, so can count more easily..

lifeline=matrix( c(sort(lifeline),seq(n.persons,1,-1)) ,nrow=n.persons); lifeline;

# (repeat for another random set of individuals)


######


# CUMULATIVE INCIDENCE OF BLADDER CANCER  (miettinen paper -- cf example 31 / 31 )

id=c(8,27,110,60,90,150)/100000; id

prob.avoid.for.next.5years=(1-id/365)^(365*5) ; prob.avoid.for.next.5years

# product of these 6 conditional proabilities

prob.avoid.for.next.30.years = prod(prob.avoid.for.next.5years); prob.avoid.for.next.30.years

# cumulative incidence 

# proportion..

1 - prob.avoid.for.next.30.years

# percent..

 100*(1 - prob.avoid.for.next.30.years)


# try using smaller or larger subunits ... eg hours, so 365*24*5 hours