There is one for serial killers and there is one for pedophiles.
Are these profiles accurate? If I know someone with one of these signs, what do I need to do about it?
You should forget about it. Those “profiles” are just a statistical correlation. They are not some sort of recipe for what makes a psycho. Seriously, those profiles are almost totally useless, except insofar as they create work for psychologists and math majors who would otherwise be Supersizing your value meal at McDonalds.
Well, what is the sign?..is this person torturing animals? Starting fires? If so, they probably need some help.
Some profiles are ridiculously detailed, so the appearence of a minor quirk is probably not something to worry about.
Profiles are not very useful in the short-term (ei – my friend is torturing cats, will he be a serial killer?) only in the larger scope of things. Such is the way with any blaket statements/generalisations.
IIRC profiling is simply a tool by which the police can narrow down their search for a perp. If your friend is exhibiting one sign, it’s probably not a big deal, but if he is exhibiting several signs…uhmmm just hide your kids and animals and don’t piss him off. 
Many people fit the profile for a serial killer to the letter but never so much as think of killing other people, likewise with the profile for pedophiles and arsonists. At the end of the day, it’s still a matter of the choices we make – we aren’t all going to decide to be policemen/women, but we aren’t all going to decide to be father-rapers either.
Just my very loosely based in reality opinionated answer to your general question. YMMV.
As has been said, it’s statistical: not every serial killer or pedophiile shows all the profile signs, and lots of ordinary people show one or more signs. The main use of profiles is to allow detectives to focus their investigations on the most likely suspects. A very high corrolation profile might indicate that the investigators are on the right track.
Another thing to keep in mind, based on John Douglas’s book, is that the profile usually comes into play after the crime has been committed. What he repeats in “Mind Hunter” is “to understand the artist, look at his work.” They look at the killer’s work, and can “profile” the killer based on that. For instance, does the evidence suggest an organized or disorganized killer? What race/sex was the victim? Was there sexual abuse, such as rape, or did the killer masturbate instead of penetrate? If the victim is strangled, did it happen from the front or the back? Was there mutilation? Anything taken?
All of that can add up to some seriously scary accurate profiles; a white male in his late 20’s or early 30’s that drives a high-mileage truck or van, stutters (but loses the stutter during the top moments of his control of the victim), is a police buff, probably has severe acne scarring, and works in construction or plumbing. That particular profile would fit a certain crime scene, but not others.
He also makes a big deal to distinguish between MO and what he calls “signature.” MO is dynamic (the killer learns from his first victims and improves over time) but signature is absolute…whatever gets the killer off will always get him off (hearing his victim scream, for example.)
I’m about 3/4 through his book and it’s really fascinating stuff. Of course, I’m sure a lot of it is “blowing his own horn” so to speak, but there are enough agencies across the country that are grateful for his work.
To answer the OP, nothing. Do nothing.
Psychological profiling is largely a media/Hollywood myth, and what little isn’t a myth is a non-science that makes no reliable and falsifiable predictions whatsoever.
Here’s some stupid reasoning: “You know all those convicted killers on Death Row? Research has found that a very high percentage of them used to eat potatoes as a child. I know this young guy who eats potatoes… oooh, maybe he’s likely to become a killer. Better be careful!”
It’s stupid, right? Okay, well, psychological profiling as presented in the popular media is on the same level. It’s as dmub as a box of hair, and it’s amazing that anyone in the news media ever bothers to give it any air-time. Or then again, maybe it isn’t amazing. Guess they have to fill the air-time with something, and interviewing a ‘profiler’ is a cheap way to take up 60 seconds.
If you want to look into it more, look up ‘false correlation fallacy’, or the logical error known by the fancy Latin name of ‘post hoc, ergo prompter hoc’.