You can sorta meet the burn-in half way with a smaller number of thermocouples (temperature is the primary component of burn-in AFAIK) and a wear model related to cumulative energy input. I've been looking to prototype this, but I've not done the work to acquire a panel which I could prototype it on. It'd take quite a bit of fast memory, but not an outlandish amount.

I hope I get a chance to take a poke at it before too many patents are filed.