How smartphones could prevent drunk driving
Could your smartphone help prevent drunk driving accidents? Could it determine if you're too intoxicated to get behind the wheel? Does it actually know when you're in a bar?
Every year in the US, some 10,000 people die as a result of car crashes involving intoxicated drivers. And聽New Year's Day is traditionally one of the worst times聽when it comes to聽alcohol-fueled wrecks.
But Israeli researchers聽have聽聽designed to work with smartphones, smartwatches,聽fitness trackers,聽and even Google Glass聽that could聽potentially聽keep intoxicated drivers off the road and save lives.聽
"Alcohol distinctly affects movement, gait, and balance in ways that can be detected by the built-in motion sensors on devices people carry around with them all the time," says Ben Nassi, a graduate student at Ben-Gurion University of the Negev, who developed the method under the guidance of Professors Yuval Elovici and Lior Rokach.
Law enforcement and advocacy groups such as Mothers Against Drunk Driving have mounted countless campaigns that warn about the dangers of drinking and driving, but technology designed to prevent intoxicated driving hasn't been widely adopted by automakers or consumers.
While drivers can聽test their own blood-alcohol level before driving by using聽a聽,聽these cost $100 to $150 and vary in accuracy. Some courts in the US聽聽to聽install an ignition control聽breathalyzer on their car that checks their聽alcohol levels before they can drive.聽But these won't help someone who hasn't been convicted yet.
Mr. Nassi says that using technology built into smart devices can聽especially聽help聽聽鈥� motorists in their 20s and early 30s, who are also most likely to聽carry or wear mobile devices wherever they go.
He聽says his聽system聽can detect illegal intoxication with 93 percent accuracy by measuring changes in gait using data collected by motion sensors built into聽the聽devices.聽The聽technique can effectively gauge intoxication with just 16 seconds of data collected as a person walks.聽
The data gets sent to a user's mobile device where an app created by Nassi's team determines the user's sobriety.聽To measure the accuracy of their method, the聽researchers verified their results against an actual breathalyzer that police departments use in Britain and elsewhere. Only one test subject was detected as sober when he was intoxicated.聽
Nassi and his team tested the technology on some聽30 patrons, male and female and most of them in their 20s,聽whom they聽encountered randomly at three bars in Tel Aviv.聽
The threshold and counting method for measuring sobriety varies among countries.聽In the US, generally the threshold for illegal blood alcohol concentration level, or BAC, is .08 or higher.聽The researchers used Israel's intoxication limit 鈥� which is 240 micrograms of alcohol per one liter of breath 鈥� as the measuring point for their experiment.聽
They asked participants to wear four Android devices 鈥� Google Glass spectacles, an LG G-watch on their left wrist, a Microsoft Band fitness tracker on their right wrist, and for those wearing pants, to carry a Samsung Galaxy S4 smartphone in their right rear pocket. Since the researchers didn't know which body movements would be most indicative of intoxication, they wanted to measure movement of the head, hands, and posterior.
They ultimately didn't need data from all four devices, however. Data from just a smartphone and smartwatch was sufficient to accurately indicate intoxication with the same degree of accuracy as the four devices together.
All of the devices except one have聽聽鈥� an accelerometer, a gyroscope, linear acceleration, gravity, and compass.聽The fitness tracker has only an accelerometer and gyroscope. The devices also all have聽Bluetooth capability to send data from聽their聽sensors to聽the聽application the researchers created and installed on the smartphone. Researchers also installed a small app on each wearable to collect and send the sensor data.聽
If Nassi's technology is adopted by phone makers or integrated into聽a third-party聽app, it wouldn't require users to enable the sobriety tester for it聽to聽work.聽Instead, it聽can be configured to automatically collect聽sensor data when the GPS feature on wearable devices聽detects聽movement. This way it's not reliant on a user who may be聽too聽intoxicated to remember to take a sensor reading.聽
The system can be refined to only take聽automatic聽readings when the user appears to be going to a bar, Nassi says.聽"Using Google maps and Foursquare it could detect that you're driving and whether there is a bar in the area [when you stop]. This is not a complicated problem."聽
The technology聽could聽also聽be designed to communicate with connected cars to prevent vehicles from starting if the data from sensors determines the device-wearer is drunk.
The research聽Nassi's team聽conducted聽builds on聽previous work聽done聽by other researchers聽showing that a person can be聽聽using the accelerometer and sensors in smartphones and other devices, as well as聽research showing that sensors in Google Glass聽聽based on head movements.聽聽
But there are some obvious privacy concerns around聽the kind of聽application Nassi's team designed. If聽insurance companies or employers can get access to the data from the sobriety app, they could聽use it to聽determine how often someone聽gets intoxicated聽on the job or in their private life.聽
"If you're turning your watch and your phone into a constant breathalyzer, if you want to collect data about your own body and use it to analyze your own physiological state, on the surface there's no problem there," says Jay Stanley, senior policy analyst with the American Civil Liberties Union's Speech, Privacy, and Technology Project.
"The only question comes if you lose control of that information," he says. "This is a reminder just how much information can potentially be teased out of rich data sources like body sensors [and] raises the stakes about the overarching privacy questions around the data that our phone has."
Nassi says privacy concerns can be addressed聽in part by encrypting the data so that unauthorized users can't read it. But if an employer or insurance company requires access to the data as a condition of employment or an insurance policy, then there would be no protection for that.
