My first thought was to look in Roark's Formulas for Stress & Strain but the condition for 4 corners + point load isn't in there. I did find this website that offers the calculations (have not vetted it though):
We get back to the old wobbly table scenario. If one of the legs is a different length than the others and the slab is assumed to be rigid, then only three supports will be engaged. However, if we allow the slab to flex, then the problem can be solved close enough for all practical purposes.
You can do a "hand-FEM", i.e., Galerkin or Ritz method with a suitable trial function. Finding a suitable trial deflection function is not necessarily trivial, though. A deflection trial for simple supports along all edges with a point load in the middle should be easy enough to find with google, and then it's just a matter of modifying it a bit and plugging into the DE (Galerkin) or functional (Ritz).