Well, the argument here is that for example, there are types of cancer which have extremely low survivalability rate, like 1 out 10 people who get it survive it. But we still treat them, and sure, right now 90% of them die but the data is used to create new drugs which in the future might improve the odds. If we just straight up said "it's very unlikely this person will survive so we won't spend $$$ on trying treatment" then we wouldn't get anywhere with many different medicines.