Abstract
Recent marketing of diabetes alert dogs (DADs) suggests that dogs
may detect an olfactory cue signaling hypoglycemia, but empirical
studies are few and their results inconsistent.We aimed to determine
whether dogs without previous sniffer dog training could be
trained to detect hypoglycemia from breath samples of T1D subjects.
Four volunteer dogs were first trained to discriminate between
breath samples from individuals without diabetes. Then breath
samples from people with T1D obtained during hypoglycemia,
normoglycemia, and hyperglycemia were presented to the dogs in
three tests. 1)Four dogs were tested for their ability to discriminate
between breath samples from the same person in different glycemic
states. 2)Two dogs were tested by presenting them with
samples from the same individual at different time points to assess
their ability to generalize the odour of hypoglycemia. 3)One dog was
tested on her ability to generalize across different people by presenting
her with sample sets from different people.
All 4 dogs successfully discriminated between hypoglycemic and
non-hypoglycemic samples (Accuracy 93.3%-100%). One dog (of 2
tested) identified hypoglycemic samples from 1 individual obtained
at different times (Specificity 89%; Sensitivity 62%). Results were
inconclusive as to whether the same dog could generalize the odour
of hypoglycemia across samples from different people.
This suggests that some dogs can be trained to detect hypoglycemic
breath of an individual with T1D but that performance may not
transfer to other individuals. Furtherwork should focus on this aspect
since it impacts training of commercial DADs.
may detect an olfactory cue signaling hypoglycemia, but empirical
studies are few and their results inconsistent.We aimed to determine
whether dogs without previous sniffer dog training could be
trained to detect hypoglycemia from breath samples of T1D subjects.
Four volunteer dogs were first trained to discriminate between
breath samples from individuals without diabetes. Then breath
samples from people with T1D obtained during hypoglycemia,
normoglycemia, and hyperglycemia were presented to the dogs in
three tests. 1)Four dogs were tested for their ability to discriminate
between breath samples from the same person in different glycemic
states. 2)Two dogs were tested by presenting them with
samples from the same individual at different time points to assess
their ability to generalize the odour of hypoglycemia. 3)One dog was
tested on her ability to generalize across different people by presenting
her with sample sets from different people.
All 4 dogs successfully discriminated between hypoglycemic and
non-hypoglycemic samples (Accuracy 93.3%-100%). One dog (of 2
tested) identified hypoglycemic samples from 1 individual obtained
at different times (Specificity 89%; Sensitivity 62%). Results were
inconclusive as to whether the same dog could generalize the odour
of hypoglycemia across samples from different people.
This suggests that some dogs can be trained to detect hypoglycemic
breath of an individual with T1D but that performance may not
transfer to other individuals. Furtherwork should focus on this aspect
since it impacts training of commercial DADs.
Original language | English |
---|---|
Article number | S49 |
Journal | Canadian Journal of Diabetes |
Volume | 41 |
Issue number | 5, Supplement |
DOIs | |
Publication status | Published - Oct 2017 |
Event | Diabetes Canada: 20th Annual Professional Conference - Edmonton, Canada Duration: 01 Nov 2017 → 04 Nov 2017 |