RadMan,
BurntKooshie is correct, but in a general way, and no criticism of BurntKooshie intended, ok to generalize theory, be brief. But doesn't demo problems re "practice", so here's the detailed version for you to evaluate before you consider research/application.
Temp rise from air entering hs/fan to some "average temp" inside the cpu = (power dissipation of cpu in watts)*(thermal resistance of hs + therm res cpu to hs interface material "adjusted for chip surface area" + therm res of cpu materials from core to case top)....therm res in degC/watt. (Ignores minor heat flow thru chip bottom/sides, as even chip mfrs do).
Look at all the "hard/?impossible? to determine" information you need in order to estimate chip's internal temp. Be happy with knowledge of the process, and measure temp of thermal diode inside Intel chips, if you have one. Even with its possible error, (admitted by Intel), you'll be much closer.
John C.