I can think of several ways to do this. Here's an example based on the built-in Missing Data Pattern command.
dt = Current Data Table();
names = dt << Get Column Names();
dt1 = dt << Missing Data Pattern(columns(Eval(names)), invisible);
n = N Row(dt1);
For(i = N Col(dt), i >= 1, i--,
If(Col Sum(Column(dt1, i + 3)) < n,
Remove From(names, i)
)
);
Try(dt << delete columns(names));
Close(dt1, nosave);