Longevity Briefs: Humans Train Dogs To Train AI To Detect Cancer

Longevity briefs provides a short summary of novel research in biology, medicine, or biotechnology that caught the attention of our researchers in Oxford, due to its potential to improve our health, wellbeing, and longevity.

Why is this research important: Dogs have a remarkable ability to recognise diseases like cancers and COVID-19 by smell, often with more accuracy than conventional testing methods. In the case of prostate cancer, dogs are even able to distinguish the benign from the more aggressive and lethal forms. Humans have made ‘artificial noses’ that are far more sensitive than dog noses. However, smell is not one molecule, but a complex pattern, much like most sounds in our environment are made up of many frequencies. Scientists are able to recognise individual molecules, but only dogs are able to recognise the patterns associated with cancer.

What did the researchers do: In this study, researchers developed a minituarised detection system that incorporates mammalian olfactory receptors that act as sensors, resulting in a device 200 times more sensitive than a dog’s nose. Data from dogs trained to detect prostate cancer was used to ‘train’ a machine learning algorithm to recognise chemical patterns associated with the disease. They then used 50 urine samples from prostate cancer patients and healthy controls to test the accuracy of both dogs and the minituarised detector.

Key takeaway(s) from this research: The success rate of the detector was similar to that of the dogs, with both methods showing over 70% sensitivity and specificity. This compares favourably to the most commonly used prostate cancer test, the prostate specific antigen blood test (PSA). According to researchers, these results provide a solid framework on which to improve this technology to a level at which it could be applied clinically. The test is non-invasive and could easily be expanded for use in detecting other forms of cancer. The detector is also small enough that it could one day be incorporated into smartphones for continuous health monitoring.

Recommended Products