The input impedance is graphed in the data sheet, but the imaginary part does not have a sign.
The impedance seems to fall gently from ~1 GHz and up. I may be possible to design a match to make it more flat, I need a 0.7-2.7Ghz flatter response. I now looks to have some 5dB drop off at 2.7Ghz.
If you could supply some s-parameter file this would be easier. A measurement on the EVAL board may be enough.