I can't read the article , but there is a explanation in "Revisiting Impulsivity in Suicide" where they point out that even if they had a 99% accurate tool measuring patients that would commit suicide. And hospitalise all of them. Because so few people are committing suicide means then same time that they would put 5 true cases out of a 100 in hospital. The other 95 shouldn't be there... So as someone mentioned earlier, it looks like it's not possible at the moment to do...