Unstratified design
Assume there are $M$ pieces of HRs. For a certain analysis $k$, the blinded estimation of the treatment effect is
$\sum_m^M HR_m \times \frac{\text{events for the m-th interval at analysis k} }{ \text{events at analysis k} }$
Stratified design
There are $S$ strata. For stratum $s$, its HR is piecewise, i.e., $HR_{s,1}, HR_{s,2}, ... HR_{s,M}$.The blinded estimation of the treatment effect is
$\sum_s^S \sum_m^M HR_{s,m} \times \frac{\text{events for the m-th interval at analysis k and stratum s} }{ \text{events across strata at analysis k} }$
We will update gs_update_ahr to enable the blinded estimation of treatment effect under a stratified design.
