B1 out 0 V=V(in) FREQ(1,0,0,5,8,45,10,18,0)
V1 in 0 ac 1
.ac dec 100 1 10
.end

The FREQ function is totally undocumented on LTspice Help file (and naively I thought also Qspice did in this way), but it’s simple to model complex impedances from large data (like this great tutorial teach me).
I’ve tried (also here naively) with TABLE(x,a,b,...) and LAPLACE on B sources, but I don’t know how to model arbitrary data with equation.

If someone could help increase my knowledge on Qspice I will really appreciate that!

This is the method I can think of if you really need R+jX type of impedance expression for .ac analysis
The idea is to use Rs-Ls model for complex impedance Z = Rs+j2pifreq*Ls. User input R and X to calculate Rs and Ls (you will get negative Ls for negative X, in theory/simulation is ok). I add attribute Rpar=0 to Ls, to force Rpar to infinity to eliminate any error introduce in calculation at extreme frequency. This demo also use a table for X value.

I output the plot of re(Z) and im(Z). You better to change y-axis to linear (de-select log). As Waveform Viewer not support Cartesian plot yet, it just plot magnitude and angle. You have to use angle to determine it is +ve or -ve. For example, at 1MHz, im(Z) reads magnitude 10 and phase 180 degree, it represent im(Z) = -10.

Post with picture cannot be edit, correct a typo. @100Hz it read im(Z) magnitude 10 and phase 180 degree (not 1MHz), that represent im(Z)=-10 @1MHz, magnitude 10 and phase 0 degree, represent im(Z)=10
that what the table do to X

You can also use Laplace G-source to simulate complex impedance. This is same simulation as above but with Laplace G-source to model complex impedance. I put a text explanation in schematic for the derive of this method.