Yes, as long as the temperature doesn't vary by more than a few degrees, it isn't a risk, but if it varies by 10 or more degrees C, then it will start causing physical expansion or contraction [the latter in the case of idle and temperature drops].  I use MSI Afterburner, so if there is a big spike in GPU utilization [note that the sample period affects the graph, so what may have been a 0.2s gap, may show as a 1s gap on the graph], I just look up at the temperature line and see if it moved more than a "blip" meaning a couple of degrees.  If I see this happen often, and especially in relatively close succession, I stop the miner and either wait it out, or mine a different pool for a bit [sometimes, conditions are such that nearly every significant pool was having similar issues].  I have not had to leave my miners down for long before I found a place to land   I give it a day or so and then test the waters.  I watch my stales carefully, compare to the fee/donation that I am paying and the variability of the pool to determine my threshold of what is acceptable.  No mathematical analysis, but I do make educated conclusions about whether I think it is a good idea to leave my miners connected to a given pool when unattended [i.e. when I am sleeping, or in the office away from my mining computers].

That's why I call it babysitting when all hell breaks loose among multiple pools