A group of researchers has developed an olfactory sensor that is capable of being able to identifying a person’s breath. Utilized in combination with an AI, the “artificial nose” authenticated up to 20 people, with an average accuracy of more than 97%.
In the current age of technology and information, the present biometric authentication is an essential security measure to protect precious assets. Breath detection is currently in addition to the plethora of biometric data machines already have to identify us.
These methods are built on the physical distinctiveness of every individual, but they’re not foolproof. Physical traits can be replicated and even altered through trauma,” explains Chaiyanut Jirayupat, the primary researcher of the study published in Chemical Communications. “Recently, the human scent has been identified as a novel type of biometric authentication, and it’s essentially using your unique chemical makeup to verify who you are.”
Percutaneous gas is comprised of the compounds created through the skin. The theory is that machines recognize these gasses. But, the skin is not producing enough chemicals that the machines can recognize. Therefore, the team looked into the use of breath as a method to identify compounds. “There can be quite a few parts per billion or trillion of volatile compounds in the skin, while exhaled compounds can have a multiple of a million,” Jirayupat continues. “Human breath has already been used to identify whether a person has cancer, diabetes, and even COVID-19 .”
The scientists first identified substances that could be utilized to authenticate biometrics. To accomplish this, they examined breath samples of participants and discovered 28 substances that could be used as options. They then created a set of olfactory sensors with 16 channels, each of which identified a specific spectrum of compounds. The sensors’ data were fed to an AI to study the composition of each individual’s breath and create a unique profile for each individual.
This system was tested with breath samples taken from six individuals. The results were positive, and it could identify the individual with an accuracy average of 97.8 percent if the sample were expanded to include 20 individuals. The level of accuracy was maintained. “This was a diverse group of individuals of different ages, genders, and nationalities. It is encouraging to see such high precision in all cases,” Explains Takeshi Yanagida, who led the study.
The researcher, however, acknowledges that we need to continue working to improve the technology before it is available on smartphones.
“In this work, we required our subjects to fast for six hours before testing,” Yanagida concludes. “We have developed a good foundation, and the next step will be refining this technique to work independently of diet. Fortunately, our current study has shown that adding more sensors and collecting more data can overcome this hurdle.”