I have a general question on survival curves. I am using the lifelines package, although can also the scikit-survival package as well. I have a population that I track over a 50 year period. I want to break the period into two halves to understand if survival time is the second half has decreased. Some ids persist from period 1 to 2, but many do not. I have the start date and end date for each id.
The event table is straightforward: id, duration (time), and event observed (boolean).
Can I simply break the full period dataframe into two halves? That does not seem correct. I feel the start date for each id should be the same, but the end date will be its last observation in the subperiod.
Or is there another recommended way to determine if the survival curves are different for each period.