"My dog does XYZ when I have an episode. Is that a natural alert?"
As a scent detection specialist, I get asked this question ALL THE TIME from service dog handlers who are hoping for natural alerts.
Unfortunately, natural alerts aren't easy to truly determine, and we need so so much more information to know for sure.
This post will break down all of the data that you need to gather, steps to take during the process, and how to correctly interpret it in order to verify that a behavior you are seeing is truly a natural alert.
Once we have the information we need to determine alert vs random behavior, we can take the appropriate next steps. Whether that's shaping it into a more specific alert behavior OR redirecting a random behavior into something else.
It's important to note that deciding whether a behavior truly is a natural alert is incredibly important from a safety perspective. Especially with psychiatric conditions that are based around stress or conditions where abortive medication is taken, the risk for false alerts causing harm is HIGH. False alerts can happen frequently if a behavior is falsely attributed to a natural alert.
With that out of the way, let's get started!
This is a piece that is typically missed in natural alert tracking.
We need data from every medical episode that you have, regardless of what your dog does.
Specifically: whether your dog was present in the home/room, the date, the time, the severity (this can be things like heart rate or blood pressure ranges or a 1-10 scale for things like panic), and other data specific to your needs.
Some examples from my students:
If you have 20+ episodes, and your dog only has a behavior change on 2 of them, the chances of a natural alert being the cause are slim.
For example, if your suspected natural alert behavior is your dog sniffing your ear, you need to track every single time your dog sniffs your ear, regardless of whether you are having a medical episode at that time.
The data you should collect for these behaviors is: Date, time, and other external factors at play.
Some past students' external factors were things like:
Sometimes, after tracking every suspected alert behavior, you'll find that your dog is only performing that behavior when their meal has been delayed or a visitor is over.
Unfortunately, that means that the behavior is functioning in a different way, and shaping that behavior into an alert would not be helpful for your medical condition.
Other times, you'll find that your dog actually does the alert behavior frequently when you aren't having a medical episode, which means that the behavior is likely not an alert and just a coincidence in timing here and there.
If you only have 2 medical episodes per month, you may need 4-6 months worth of data to comfortably call a behavior a natural alert.
It's important to have long term data before making any steps towards shaping an alert to verify continual patterns, not just temporary ones.
I like to see at least 20 episodes with behavior changes, less than 2-3 missed episodes, and no behavior changes outside of medical episodes before calling something a natural alert.
Rewarding any behavior will cause them to increase in frequency.
We want data of what happens naturally, and adding rewards can affect that.
If your dog stops the suspected natural alert behavior after a few weeks of tracking, unfortunately it was likely not a natural alert, and another method of training will be a better route for your pup.
This is a HUGE one.
So often I'll hear from students that they've been tracking for weeks, but when I ask what the data from the tracking shows, they let me know that they've only been loosely keeping track in their head, and they don't have concrete numbers.
Unfortunately, our brains will throw out data that seems unimportant, like a dog whining at you for a meal or other times that aren't a medical episode. This is crucial for determining whether a behavior is a true natural alert.
It's not enough to just mentally keep track. Use a pen and paper, or even a spreadsheet if that's your thing. Data is your friend here!
Along with data tracking, I would be silly not to include a small section on a common occurrence with my students.
Usually dogs who are responding to their own stress vs naturally alerting show typical signs of:
These behaviors commonly happen at the onset of the episode vs before, but some dogs learn to predict them and the behaviors leak in prior to the episode onset.
While this may be seen as a form of a natural alert, it's important to consider the dog's experience before shaping a natural alert.
If your pup is this stressed by a medical episode, training energy is best spent working with them to reduce their stress instead of bringing more attention to the medical episode with alert training.
The best natural alert dogs are not bothered by the medical episode, just noticing them.
Natural alerts are rare, but they do happen.
Tracking this data and being careful about potential stress are two ways to make sure that what you are seeing really lines up with a natural alert.
From a safety perspective, false alerts can be incredibly detrimental and in some cases even increase episodes in their handlers, which is why ensuring a true natural alert is so important.
If you have a potential natural alerting dog, and you want some help determining next steps, you'd love my service dog membership the Pack. Inside, you can get my review of your data as well as recommendations for shaping a natural alert for a lower cost than private lessons.
And if you realize that you don't have a natural alert, but you have a behavioral sign that could clue your dog into an episode, you can use my free webinar to teach a behavior based alert here: https://www.deltatails.com/df-behavior-interruption
Weekly emails with tips, tricks, and motivation to train your dog to listen to every cue, joyfully.