Well, you can certainly ballpark it pretty easily. This reaction entails forming an O-H and C-H bond and breaking a C-O and H-H bond, and releasing any angle strain energy involved in the epoxide. Although bond energies vary, I think it's ok to assume that taking average values will give us an estimate within about 10% of the real value. O-H bonds are about 110 kcal/mol and the C-H of a benzylic hydrogen (C6H5C-H) is about 113 kcal/mol, slightly more energetic than is typical for alkane C-H bonds (around 100 kcal/mol). A quick check around the internet puts angle strain values for ethylene oxide and 3-member alkane rings at 13 and 27.5 kcal/mole, respectively. So, summing up all these values gives a back of the envelope estimate of reaction enthalpy to be 188 kcal mole absorbed due to bond breaking, 223 kcal/mol released due to bond formation, and an additional 13-28 kcal/mole released due to bond strain. Allowing for extra wiggle room, I'd put a rough estimate of reaction enthalpy to be around 45-65 kcal/mol released, depending on what the energy content of the strained epoxide ring is. Seems to be in line with the patent and article you mentioned first.
Of course, a calorimetric measurement or even a good computer calculation will be a better guide.