survexp.fit {survival} | R Documentation |
Compute expected survival times.
survexp.fit(group, x, y, times, death, ratetable)
group |
if there are multiple survival curves this identifies the group, otherwise it is a constant. Must be an integer. |
x |
A matrix whose columns match the dimensions of the |
y |
the follow up time for each subject. |
times |
the vector of times at which a result will be computed. |
death |
a logical value, if |
ratetable |
a rate table, such as |
For conditional survival y
must be the time of last follow-up or death for
each subject.
For cohort survival it must be the potential censoring time for
each subject, ignoring death.
For an exact estimate times
should be a superset of y
, so that each
subject at risk is at risk for the entire sub-interval of time.
For a large data set, however, this can use an inordinate amount of
storage and/or compute time. If the times
spacing is more coarse than
this, an actuarial approximation is used which should, however, be extremely
accurate as long as all of the returned values are > .99.
For a subgroup of size 1 and times
> y
,
the conditional method reduces to exp(-h) where
h is the expected cumulative hazard for the subject over his/her
observation time. This is used to compute individual expected survival.
A list containing the number of subjects and the expected survival(s) at each time point. If there are multiple groups, these will be matrices with one column per group.
Most users will call the higher level routine survexp
.
Consequently, this function has very few error checks on its input arguments.