An app that knows when you've been in an accident - and then calls emergency services for you

Washington

IMAGINE you're riding a bicycle when you collide with a car and are thrown over the handlebars.

When you hit the street, you're knocked unconscious and unable to call for help. In this scenario, you're wholly dependent upon a stranger coming to your rescue.

If you're in public, surrounded by people, that might not be a problem. But if you're injured when you're alone, hours could pass before you're found.

The makers of one app claim that their technology now includes a feature they refer to as "automatic crash detection and response".

As of last week, Noonlight - an app designed to connect people with rescue services - began using an algorithm that taps into a smartphone's sensors, allowing the platform to measure and detect minuscule changes in the user's location, motion and force.

If the sensors detect a sudden change in motion and force indicating that the user has been involved in an accident, the app alerts 911 without the user's prompting.

The company will charge users about US$10 a month for its crash-detection services.

"The platform is agnostic to the vehicle you're in," said Noonlight co-founder Nick Droege, who believes we are moving increasingly close to a time when our smartphones and virtual assistants will perpetually monitor our health and vital signs.

"You don't even have to be driving a vehicle for this platform to work. You could also be on a bicycle or a Bird or Lime scooter, and the app will know whether you've been in an accident."

Automatic crash-detection devices may be new for smartphones, but they're hardly novel. For years, automakers have offered systems such as OnStar, a subscription service that can connect drivers with emergency workers and includes crash detection. BMW, Toyota and Ford offer systems that connect drivers with emergency services, though some are not automatic and require the push of a button.

A study released by the US National Highway Traffic Safety Administration found that automatic collision notification has the ability to "dramatically" reduce emergency notification times after car accidents, particularly in rural areas.

"Reductions in rural average crash notification times from nine minutes to one minute after the crash have been estimated to potentially save 3,000 lives per year," the study noted.

For crash detection to work accurately via a smartphone, Mr Droege said, the system relies on the device's GPS, accelerometer and gyroscope, both of which are used for maintaining orientation.

The app might also tap into sensors such as a phone's proximity meter or magnetometer. The former measures a phone's proximity to another object and the latter provides the phone with orientation to the earth's magnetic field.

The sensors are used in conjunction with analytics data from billions of miles of driver data from Zendrive, a company that amasses data about driver behaviour, such as aggressive and distracted driving. An algorithm pores over the dual streams of information, Mr Droege said, noting that the data is regularly updated in hopes of improving the algorithm's accuracy.

What if a crash has occurred but emergency services aren't needed?

When the app thinks a crash has occurred, Mr Droege said, the platform sends out a virtual alarm of sorts, one that contacts emergency responders and shares the user's location and relevant medical information (such as allergies and medication use).

At the same time, he added, users receive a text from the company's dispatchers asking them whether they need help. A user can cancel an active alarm by sharing a four-digit code. The entire process occurs in about two minutes in most cases.

"Our big audience is potentially parents who can add this to a teenager's phone so they know that in the event that something happens on the road, help will get to them," Mr Droege said.

"We also think it can help with the elderly, who may have trouble calling emergency services." WP