Particle-to-particle hydrodynamic force and torque  fluctuations is a well documented feature of particle-laden flows.  Classical correlations ignore these fluctuations and only require the  knowledge of the locally averaged solid volume fraction and Reynolds  number to estimate the hydrodynamic momentum transfer (primarily the  drag force). However, in the case of the drag force, the magnitude of these fluctuations can be as large as the average value, thus preventing  coarse-grained numerical models of particle-laden flows from delivering  high fidelity predictions. Recent advances on this problem involve taking advantage of the local microstructure surrounding each individual  particle and using a description of this microstructure as an additional input parameter for deterministic hydrodynamic drag, lift and torque predictions. Modelling the functional dependence of the  hydrodynamic force and torque on the microstructure constitutes a  significant challenge, and so far only the reverse problem of the flow  past a random array of stationary particles has been thoroughly  investigated in the literature. I will discuss various so-called microstructure informed models of hydrodynamic force and torque  fluctuations in a random array of particles that we recently proposed  and the challenges lying ahead to provide the multiphase flow community with high fidelity and deterministic hydrodynamic force and torque models that are compatible with the coarse-grained two-way  Euler-Lagrange formalism.