One day, a machine will smell whether you're sick

Modern medical research confirms smell of skin, breath, bodily fluids can be suggestive of illness

Blindfolded, would you know the smell of your mom, a lover or a co-worker? Not the smells of their colognes or perfumes, not of the laundry detergents they use — the smells of them?
 
Each of us has a unique “odourprint” made up of thousands of organic compounds. These molecules offer a whiff of who we are, revealing age, genetics, lifestyle, hometown — even metabolic processes that underlie our health. Ancient Greek and Chinese medical practitioners used a patient’s scent to make diagnoses. Modern medical research, too, confirms that the smell of someone’s skin, breath and bodily fluids can be suggestive of illness. The breath of diabetics sometimes smells of rotten apples, experts report; the skin of typhoid patients, like baking bread.

 
But not every physician’s nose is a precision instrument, and dogs, while adept at sniffing out cancer, get distracted. So researchers have been trying for decades to figure out how to build an inexpensive odour sensor for quick, reliable and noninvasive diagnoses.
 

The field finally seems on the cusp of succeeding. “You’re seeing a convergence of technology now, so we can actually run large-scale clinical studies to get the data to prove odour analysis has real utility,” said Billy Boyle, co-founder and president of operations at Owlstone, a manufacturer of chemical sensors in Cambridge, England.
 
Boyle, an electronics engineer, formed the company with two friends in 2004 to develop sensors to detect chemical weapons and explosives for customers, including the United States government. But when Boyle’s girlfriend and eventual wife, Kate Gross, was diagnosed with colon cancer in 2012, his focus shifted to medical sensors, with an emphasis on cancer detection.

 
Gross died at the end of 2014. That she might still be alive if her cancer had been detected earlier, Boyle said, continues to be a “big motivator.”
 
Owlstone has raised $23.5 million to put its odour analysis technology into the hands of clinicians. Moreover, Britain’s National Health Service is funding a 3,000-subject clinical trial to test Owlstone’s sensor to diagnose lung cancer.
 
The sensor is a silicon chip stacked with various metal layers and tiny gold electrodes. While it looks like your mobile phone’s SIM card, it works like a chemical filter. The molecules in an odour sample are first ionised — given a charge — and then an electric current is used to move only chemicals of diagnostic interest through the channels etched in the chip, where they can be detected. “You can program what you want to sniff out just by changing the software,” Boyle said. “We can use the device for our own trials on colorectal cancer, but it can also be used by our partners to look for other things, like irritable bowel disease.”

 
The company also is conducting a 1,400-subject trial, in collaboration with the University of Warwick, to detect colon cancer from urine samples, and is exploring whether its chips can help determine the best drugs for asthma patients by sorting through molecules in their breath. A similar diagnostic technology is being developed by an Israeli chemical engineer, Hossam Haick, who was also touched by cancer.
 
“My college roommate had leukemia, and it made me want to see whether a sensor could be used for treatment,” said Mr. Haick, a professor at Technion-Israel Institute of Technology in Haifa. “But then I realized early diagnosis could be as important as treatment itself.”

 
His smelling machine uses an array of sensors composed of gold nanoparticles or carbon nanotubes. They are coated with ligands, molecular receptors that have a high affinity for certain biomarkers of disease found in exhaled breath. Once these biomarkers latch onto the ligands, the nanoparticles and nanotubes swell or shrink, changing how long it takes for an electrical charge to pass between them. This gain or loss in conductivity is translated into a diagnosis.
 
“We send all the signals to a computer, and it will translate the odour into a signature that connects it to the disease we exposed to it,” Haick said. With artificial intelligence, he said, the machine becomes better at diagnosing with each exposure. Rather than detecting specific molecules that suggest disease, however, Haick’s machine sniffs out the overall chemical stew that makes up an odour. It’s analogous to smelling an orange: Your brain doesn’t distinguish among the chemicals that make up that odour. Instead, you smell the totality, and your brain recognises all of it as an orange. Haick and his colleagues published a paper in ACS Nano last December showing that his artificially intelligent nanoarray could distinguish among 17 different diseases with up to 86 per cent accuracy.