hexomino is correct that there's not a safe way to proceed, but the heuristic doesn't give the right probabilities for the remaining squares.
Looking at the top five unknown squares, there are five possibilities for that region, each of which corresponds to exactly four solutions for the whole grid. These are (x=mine, o=empty):
oo xo ox oo xo
xx xx xo ox ox
o o x x x
This gives probabilities (left to right, top to bottom) of 2/5, 1/5, 3/5, 4/5, 3/5.
However, that's not really relevant.
The four squares in the middle contain two squares along a diagonal, and the two squares to the bottom of the group of seven contain exactly one mine. There will never be a way to figure out these six squares with more than 50% chance even if you solve the rest of the mine.
However, guessing one of those groups can solve both (if you guess the right way), and also tell you about the bottom square in the group of five. So the right thing to do now (in terms of maximising your probability of clearing the grid) is guess that the two squares at the bottom of the group of seven have a mine on the left-hand side and an empty square on the right. If this works, you have got information about the four squares diagonally below and the one immediately above. Since you can't avoid this 50% guess, you haven't lost anything if it doesn't work.
This strategy gives a 40% chance of winning. If you try the "safest" square first, even if it's not a mine you still won't be able to distinguish between the first and last configurations above, and you'll still have to guess in the other six squares, meaning you can't do better than 30%.