Odor Sensors; Sniffing out the Disease in Us with Artificial Intelligence


Every human has their own unique odor made up of uncountable organic compounds. These compounds show our internal composition and reveal to people the details of who we are, our age, genetics, lifestyle and the underlying metabolic processes of our bodies.

Years before modern medicine, ancient Chinese and Greek physicians used the scent of a patient to diagnose them. Today, modern research confirms too that our body’s fluids, skin and breath can tell us a lot about an illness. It is said by experts that the breath of a patient with diabetes will smell like rotten apples whereas the patients with typhoid have skin that smells like baking bread.

But where doctors and cancer sniffing dogs are not always enough, there are other options that scientists have been working on for decades that are affordable, reliable and provide noninvasive diagnosis, like odor sensors.

It finally looks like their efforts will wage well. A manufacturer of chemical sensors from England, Billy Boyle said that all technologies are coming together so he expects that soon large scale medical studies will be run to obtain the data needed to prove the efficacy of odor analysis.

President and Co-founder at Owlstone and an electronics engineer, Boyle created his company with two of his friends in 2004 to make sensors that detect explosives and chemical weapons for clients that included the government of the United States. However, when in 2012 Mr. Boyle’s spouse Kate Gross got diagnosed with cancer, his focus shifted towards medical sensors, prominently on cancer detection.

His wife, Ms. Gross died in 2014 and according to Boyle, a big motivation for him is that if her disease had been detected earlier she could have still been alive.

The company Owlstone generated $23.5 million to give its technology of odor analysis to doctors dealing with patients.  With that, the National Health Service in Britain is investing in a clinical trial of 3000 subjects to test the company’s sensors to identify lung cancer.

Their sensor, which looks like a sim card and works like a chemical filter, has a silicon chip with layers of metal and small electrodes made of gold.

The odor sample’s molecules are ionized firstly and then electric currents move the chemicals of diagnostic interest within the channels of the chip where they can be diagnosed.

Boyle added that what needs to be sniffed out can be changed through the software enabling them to use the sensor for trials on colorectal cancer while their partners can work on conditions like the irritable bowel disease.

They are also conducting a trial of 1400 people along with the University of Warwick to diagnose colon cancer through urine samples and testing chips to check if they can help doctors determine the best medications for asthma sufferers by going through their breath molecules.

Another chemical engineer who was touched by cancer, Hossam Haick, is also working on a similar technology in Israel.

His college friend’s leukemia made him want to test whether a sensor would help in the treatment but he soon realized that an early diagnosis was as important as the treatment itself.

Haick’s sensors are made of gold nanoparticles and carbon nanotubes with ligand coating and molecular receptors which tend to detect biomarkers of illnesses found in the breath exhaled.

When these biomarkers stick to ligands the nanoparticles and tubes shrink or swell affecting the time it takes the electrical charge to flow between them. This loss or gain in conductivity is then deciphered to diagnose disease.

The signals are sent to a computer where they decode the scent to a signature which then connects it to the illness it was exposed to.

Haick stated that with artificial intelligence the machines are getting better at identifying with each exposure. Instead of recognizing specific compounds which propose the diseases, Haick’s machines smell the overall chemical composition that creates an odor.

Last December Haick and his team published their research in ACS Nano which revealed that his AI-powered non-array was able to differentiate between 17 diverse illnesses with a precision of 86%.

Teams in the US also recently were given an $815,000 grant a few months ago from the Kleberg Foundation to work on an odor sensor which would spot ovarian cancer in blood plasma. Their sensors use snippets of DNA to stick on to the odor particles.

Along with these teams, many others are working in Austria, Japan, and Switzerland to develop similar sensors for odors to diagnose illnesses.

While all the different scientists are in competition, their goal of a breakthrough is full of life-saving possibilities across the entire disease spectrum. And industry specialists believe that in about 5 years the tools should be ready for medical practitioners to use in their clinics and hospitals.