You cannot solve it to be fully compatible with future releases. You will need to stop doing that.
Or you could cheat:
e = '(-1)*(((2.1967)-(x3))*((x4)+(x2)))'
vars_char = symvar(e);
vars_sym = sym(vars_char);
vars_cell = num2cell(vars_sym);
fun_str = ['@(', strjoin(vars_char, ','), ') ', e];
fun_handle = str2func(fun_str);
result = fun_handle(vars_cell{:});
However, this will not be exactly the same, in that numeric values will be calculated out, and floating point numbers will typically be converted to rational. For example, 'sin(1) + x' would come out as x + 3789648413623927/4503599627370496
