I am suspicious.
While it’s certainly possible that the dog may be be picking up on indicators, there’s no way that it’s more accurate or effective than testing blood sugar. The whole idea behind regular self testing is to regulate the blood sugar so that the patient does not have dangerously high and low levels, which is what the article seems to indicate that the dog picks up on.
It’s my understanding that individuals suffering from diabetic ketoacidosis may have a sweetish body odor or breath, but if you’re ketoacidotic, you are sick. Alternatively, if she has very low blood sugar, she may behave more like a seizure patient, something else that the service dog may be picking up on.
I think it’s possible that the presence of the dog may be a good idea if the child (or her parents) aren’t doing an adequate job of monitoring her, but that should be an adjunct, not the norm. My guess is that the tone of the article (suggesting that the dog is the main monitor of the child’s health) is misleading. I think it more likely that the dog is a last stage warning system that something may be seriously wrong.