Given the low resolution needed, would be the shunt temperature a good enough measurement of the current?. I think that isolation is achieved easier when measuring temperatures than voltages. You can use a second temperature sensor to measure ambient temperature and substract both values to derive a rough value for the current (delta T is aprox. proportional to the power in the shunt, which is aprox. proportional to the square of the current).

Of course, the thermal time constant might be longer than you need to get a response in seconds.

Alfredo.