Comment by seanmcdirmid

They don’t exactly need a control group if the chance of living N weeks is effectively zero. They would need it for something like living N weeks and living N+1 weeks, but any kind of change from an individual expected outcome would be useful signal for maybe more rigorous testing.