Canine Signalling Interfaces for Bio-Detection Dogs

In this project, which did not did not require licensing under the Animals (Scientific Procedures) Act 1986 (ASPA), we are designing a canine-centred sensor system that allows bio-detection dogs to communicate accurately and reliably their spontaneous response when they detect the odour of human disease in biological samples such as urine, sweat or breath.

Background

Although common, some human disease, such as cancer, are difficult to diagnose. Dogs can be trained to recognise the odour of diseased cells in biological samples and clinical trials are increasingly highlighting the potential of bio-detection with dogs. However, the conventional signalling protocols the dogs are trained to use to communicate what they detect (e.g. sitting down in front of a positive sample) are inconsistent with their evolutionary response to odours of interest and only allow them to provide binary responses (‘yes’ or ‘no’), thus interfering with and limiting the potential of the dogs’ detection work.

Research conducted

We hypothesised that the dogs’ interaction with the samples would vary depending on the samples, and that it would be possible to infer more nuanced information about the samples’ content based on the dogs’ spontaneous responses to them. So, in collaboration with UK Charity Medical Detection Dogs, at the Animal-Computer Interaction Laboratory we have been developing a canine-centred interactive system to record and analyse dogs’ spontaneous interaction with odorous substances.

By enabling the dogs to spontaneously respond to the olfactory stimulus, we aim to increase signalling reliability and, by enabling them to provide nuanced responses, we aim to increase signalling accuracy.

Research findings

Our findings suggest a correlation between the patterns identifiable in the data and the content of the samples screened by the dogs. For example, we have observed variations that are proportional to the concentration of amyl acetate samples. We have also identified false positives, when the dogs’ behavioural responses erroneously indicated a positive response.

By enabling these canine workers to express what they find more reliably and accurately, this research has the potential to transform bio-detection with dogs. One of its main contributions will be to help inform the development of electronic biosensors for diagnostics on a large-scale. 

By enabling the dogs to communicate on their own terms, this technology can much better convey what they know, including nuances that may indicate different stages of disease. The key is recognising that animals are legitimate stakeholders and need to be at the centre of the design process.

Dr Clara Mancini
Senior Lecturer in Computing and CommunicationsVisit Clara's profile