Displaying 1 result from an estimated 1 matches for "frailty_i".
Did you mean:
frailty
2012 Feb 10
0
coxme with frailty
...ng easier...). They proceed in 4 steps:
1. Obtain the cumulative hazard function for each subject.
2. Choose an arbitrary value for the variance parameter of the
frailties (call it theta).
3. Compute for each subject an estimate of the value of their
frailties, USING this variance parameter theta:
frailty_i= \frac(1+\theta \times c_i}{1+\theta \times H_i} (formula on
p. 321), where H is the cumulative hazard for the subject. So if theta
is 0 (no variance), then frailty=1 (i.e., no excess frailty). As theta
goes to infinity, the estimated frailty is simply the ratio
1/(cumulative risk so far) or 1/(cum...