Recently, a fascinating research study made headlines internationally. In the September 2004 issue of the British Medical Journal, British researchers reported using trained dogs to detect bladder cancer by sniffing human urine, opening up the possibility that dogs may one day be used to detect the disease.
What the media did not report is that the dogs in this research project were clicker trained. We have also heard directly from a researcher at a West Coast university, where a similar project is underway—in which they are also clicker training the dogs.
The marker signal makes it easy for the dogs to understand what question they are being asked. Instead of trying to please the trainer, or just following orders to stay out of trouble, they are able to focus on the exact job: is that particular chemical combo, the one they get clicked for, present or absent?
The news reports I've seen made no mention of the training procedure, nor did the medical commentators; but I can't help but think that this operant response is going to prove to be even more useful in the future, because it can be very accurate. The investigators were surprised that several dogs identified as cancerous samples from one person who had been selected as a control, a person with a clean bill of health, but who indeed proved to have a kidney malignancy.
On the other hand, some positive samples were not identified by every dog. Rather than assume the dogs are not after all very reliable, I wonder if it's possible that there are different kinds of cancers involved, with different chemical signatures. In that case, some dogs might just need more information (i.e., "Look for scent A and also for scent A+ and B"). The clarity of the clicker communication should make it possible to refine the task in many ways, and at the same time to be able to rely on the dogs' accuracy with increasing confidence.
The original research paper, "Olfactory detection of human bladder cancer by dogs: proof of principle study," is available online at the British Medical Journal website. Click here to read the article.
Post new comment