You can't deduce airspeed from the data you are suggesting, though.
At least, not with any expectation of accuracy.
A wing flexes the same under load to support the same weight, independent of airspeed (to a first approximation). So the wing shape at 100knots and the wing shape at 200knots would be essentially identical for the same weight. Even then, backing out the weight would not be trivial, especially given the unknowns of the inertia distribution due to fuel and payload distribution. (The wing flex for a heavy a/c with low fule state will differ from that with a high fuel load but correspondingly lower payload)
So, let's assume nevertheless that somehow we back out a reliable estimate for the weight. What's left in the lift equation - 1/2 rho V squared, S and CL. OK, S we can define, but to get closer to V we need CL. For that we're going to need an accurate model of the aerodynamics - not at all trivial, especially for an in-service aircraft not given the TLC a test vehicle gets. (Indeed, this methiod has been used with a KNOWN V to estimate degradation of CL compared to standard, and it's not always small).
I think if you do the error budgets in that process you struggle to get any kind of accuracy for airspeed - especially given that we have a pretty good way of measuring it right now that's a great deal simpler.
The alternative - Well, you could attempt to deduce the airspeed by looking at very small differences in wing loading due to low order Mach effects - not something terribly accurate at low Mach numbers. But that's even more error prone I suspect.