The findings, which researchers said could someday lead to an automated odour-detection system small enough to be incorporated into a mobile phone, have been published in PLOS One. The paper is authored by Clare Guest of Medical Detection Dogs in the UK, research scientist Andreas Mershin of MIT, and 18 others from various universities and organisations including Johns Hopkins University and the Prostate Cancer Foundation.
Studies have shown that trained dogs can detect diseases including various types of cancer, and potentially COVID-19. In some cases, involving prostate cancer for example, dogs had a 99 per cent success rate in detecting the disease by sniffing patients’ urine samples.
Training dogs is time-consuming, and scientists have been looking for ways to automate the olfactory capabilities of the canine nose and brain in a compact device. The system developed by the team at MIT and other institutions has been coupled to a machine-learning process that can identify distinctive characteristics of the disease-bearing samples.
Mershin explained that dogs have been shown to be some of the earliest, most accurate disease detectors with performances in controlled tests exceeding that of the best current lab tests. Furthermore, dogs pick up connections that have so far eluded human researchers: when trained to respond to samples from patients with one type of cancer, some dogs have identified several other types, even though similarities were not evident to humans.
According to the team, its miniaturised detector system incorporates mammalian olfactory receptors stabilised to act as sensors whose data streams can be handled in real-time by a typical smartphone. Such detectors could potentially pick up early signs of disease far sooner than typical screening regimes, Mershin said, as well as potentially warning of smoke or a gas leak.
The team tested 50 samples of urine from confirmed cases of prostate cancer and controls known to be free of the disease, using dogs trained and handled by Medical Detection Dogs in the UK and the detection system itself. When applied to a machine learning program, the artificial system was said to match the success rates of the dogs with both methods scoring over 70 per cent.
Mershin said that the miniaturised system is actually 200 times more sensitive than a dog’s nose in terms of being able to detect and identify tiny traces of different molecules, as confirmed by controlled tests mandated by DARPA - but the system requires the addition of machine learning to help to interpret those molecules and find the patterns that dogs can infer from the scent.
Researchers said the achievement provides a solid framework for the development of the technology to a level suitable for clinical use. Mershin now hopes to test a larger set of samples to pinpoint in greater detail the indicators of disease.