Turn your phone into a breathalyzer

BGU student develops app for smartphone

Man talking on mobile phone. (photo credit: INGIMAGE)
Man talking on mobile phone.
(photo credit: INGIMAGE)
A smartphone app that can be turned into a virtual breathalyzer to detect whether its user is inebriated has been developed by Ben-Gurion University researchers.
Ben Nasi, a master’s degree student, used a special program to create the virtual breathalyzer, using not only a smartphone but also a smart watch or a Google Glass device.
The system documents just once the walking patterns of people from their car to a bar, and compares his walking out of the bar to determine his level of drunkenness, if any.
Drunk driving is a problem all over the world.
In the US, every 51 minutes on average, a person is killed in a road accident due to inebriation.
Many attempts have been made to reduce drunk driving, said the university, but this is the first one that uses smart devices to fight it.
Prof. Yuval Alovich, Prof. Lior Rokah and Nasi of BGU’s department of software and information systems used existing sensors in smartphones to identify the bodily changes in gait as a result of drinking alcohol and learning algorithms.
They used data collected from 30 people, three fifths of them men and the rest women, who used a Galaxy S4 smartphone, LG-G smart watch or Microsoft smart bracelet. The participants were examined when going to three different Tel Aviv bars over five nights.
The results were compared with those of a “Yanshuf” breathalyzer operated by the Israel Police and found that the system identified correctly all the persons who were over the legal driving limit.
The researchers said that the system works most accurately on the smartphone and the smart watch. A total of 275 million of these devices were bought in 2016, and sales are due to reach 322 million in 2017. As the biggest group of consumers of these devices are young people, who also tend to go to bars, the technology could save a lot of lives and is relevant to them, Nasi concluded.