I'm surprised to see that this question How to roughly know if a electronic scheme will fail soon and protect from it? has been put on hold as primarily opinion-based.
It could (maybe) have been put on hold as too broad, because the OP could've narrowed more the question. But, in any case, it's clear to me that the question can be objectively answered within the framework of RAMS engineering, a very specific expertise field.
The OP was asking (maybe without knowing the exact technical terms) two things:
Is there any reliability data (MTTF, failure rate...) available for Raspberry Pi and/or similar SBCs?
How can I calculate the MTTF (mean time to failure) and then the MTTM (mean time to maintenance) so I can preemptively replace the boards before a failure?
This is not the realm of "almost entirely based on opinions, rather than facts, references, or specific expertise". It's 100% reliability engineering.
Would it be worth if a moderator corrected this situation? Or, alternatively, what changes needs the question in order to be reopened?
I was really curious about what could other people with RAMS background answer to the OP's question. I'm here to volunteer, but also to learn. That could have been a good question for someone with firmware/software RAMS background to step in. My answer was focused just in the hardware aspects of RAMS.