New machine-learning system detects diseases from air samples better than dogs

Numerous studies have shown that trained dogs can detect many types of diseases,

including lung, breast, ovarian, urinarybladder and prostate, and possibly COVID-19, just by smell. In some cases, such as those associated with prostate cancer, dogs are 99% likely to detect the disease by sniffing out urine samples from patients.

But it takes time to train such dogs.Scientists have been looking for ways to automate the amazing olfactory abilities of the canine nose and brain in a compact device. Now, a team of researchers from MIT and elsewhere have developed a system that can determine the chemical and microbial content of an air sample with even greater sensitivity than a dog's nose. Scientists have combined it with a machine learning model that identifies the distinctive characteristics of disease-signaling patterns.

Discoveries that, according to researchers,could someday lead to an automated odor detection system small enough to fit into a mobile phone, published in the magazine PLoS ONE in an article by Claire Guest of Medical DetectionDogs in the UK. Researcher Andreas Mershin from MIT and 18 other scientists from Johns Hopkins University, the Prostate Cancer Foundation and a number of other institutions and organizations also took part in the project.

Study design for testing dogs' sense of smell.(A) Two dogs, Florin and Midas, selected to participate in the trial. (B) Image of presentation containers. (C) The test containers are placed on a metal arm attached to the carousel. (D) Comparison of indications for biopsy-negative control and cancer specimens in a double-blind study. This table shows that out of 21 controls, Florin gave 5 false positives, resulting in 76.2% specificity compared to 6 false positives from Midas, resulting in 70% specificity. Both dogs correctly identified 5 of the 7 target samples giving a sensitivity of 71.4. Credit: PLOS ONE, CC-BY 4.0,

Over the past few years, scientists have been improvinga miniature detector system that includes mammalian olfactory receptors stabilized to act as sensors. Data streams are processed in real time using the capabilities of a conventional smartphone. The detectors, equipped with advanced algorithms developed through machine learning, could potentially detect early signs of disease much earlier than conventional screening, the scientists said.

In the last tests, the team tested 50 samplesurine from confirmed prostate cancer cases and a healthy control group using dogs trained and maintained by Medical Detection Dogs in the UK and a miniature detection system. They then applied a machine learning program to identify any similarities and differences between the samples that could help the sensory system to identify the disease. When testing the same samples, the artificial system was able to compare the success rates of dogs, with both methods scoring over 70%.

According to scientists, the miniature detection systemis actually 200 times more sensitive than a dog's nose in terms of the ability to detect and identify tiny traces of various molecules. This has already been confirmed by controlled tests authorized by DARPA. Scientists used machine learning to correctly interpret the molecules. Their goal is to try to find subtle patterns that dogs can deduce from smell, but humans cannot understand through chemical analysis.

Read more

The first accurate map of the world was created. What's wrong with everyone else?

What drones are there and when will passenger air taxis appear?

Check out the most beautiful pictures of Hubble. What has the telescope seen in 30 years?

DARPA - Office of AdvancedUS Department of Defense Research Projects - A department of the US Department of Defense responsible for developing new technologies for use in the interests of the military.