Catching Drivers Who Are Stoned Or Tired Could Be Like a Game

An In-depth Article on Impairment Detection Technology by Auto Insurance Report

The technology to detect sleepy, stoned or drunk drivers could be in the palm of your hand, if startups are successful at marketing gamelike smartphone applications that detect impairment.

Two of the leading startups, Impairment Science and Predictive Safety, are marketing their apps to insurers and employers to identify tired or stoned workers, especially in industries with high safety stakes, such as trucking, mining, emergency services and manufacturing. Several peer- reviewed scientific studies have confirmed the effectiveness of the apps – including research at Johns Hopkins University School of Medicine using Impairment Science’s app Druid. While the technology hasn’t yet caught on with auto insurers, the app makers say carriers are showing interest.

The phone and tablet applications measure cognitive ability and motor skills by pacing users through gamelike tests that require matching, remembering specific shapes or clicking circles and squares as they briefly flash on the screen. Impairment Science’s app also requires users to balance on each leg.

The companies believe smartphone apps could be the answer to catching drivers who are impaired by marijuana, an elusive goal absent a device like the ubiquitous Breathalyzer that tests for blood alcohol levels. Apps could also simplify drug recognition evaluations (DRE) – roadside tests that police conduct with suspected drug-impaired drivers who pass Breathalyzer tests. The12-step DRE requires measurements with a pupilometer, muscle tone examinations, taking a pulse reading and other time-intensive evaluations.

“What we’ve done is we’ve taken the best elements of the standard field sobriety test – the ones that are most highly correlated with impairment over time with scientific studies – and we digitize them,” said Rob Schiller, CEO of Impairment Science. “And instead of being a three-minute standard field sobriety test or a 30-minute DRE test, we’ve turned it into one minute, and we’ve made it an objective test.”

Both Impairment Science and Predictive Safety are marketing their apps to insurance carriers, suggesting that they require truckers or other workers to take the test before work to maintain commercial auto or workers comp coverage. Beer conglomerate AB InBev, the owner of Anheuser Busch, Modelo and other brands, has tested truck drivers with Impairment Science’s app in several Central and South American countries over the last year, with negotiations to extend testing into a permanent program, according to Schiller.

Several other companies have piloted the apps to study fatigue or impairment among their workers, and Schiller said Impairment Science is currently in talks with one smaller specialty insurance writer and one larger national company. Most auto insurers, however, remain largely unaware of the technology, according to commercial insurance brokers. With persistently high loss ratios in commercial auto, the makers of impairment-detection apps see opportunity in the new wave of startup insurance companies looking to use telematics, dashcams and other new technology to reduce losses.

“We are getting a lot of interest from the insurance industry,” Predictive Safety COO Jeff Sease said in an email. “Lots of money has been invested in camera systems and advanced telematics, but now that is opening corporate eyes to the reality of fatigue on the road. ... Our approach is to address fatigue before you get behind the wheel, which becomes highly complementary.”

In 2023, the logistics company ConGlobal used Predictive Safety’s AlertMeter to test employees before shifts at 40 shipping container receiving facilities, resulting in a 20% reduction in reported injuries compared to other sites, according to a published case study. The company expanded testing to 35 more locations last year. In another case study, the Washington Metro Transit Authority regularly tested about 1,000 drivers in its paratransit division with the AlertMeter app, pulling those with bad scores from the road. The paratransit buses are equipped with cameras that are triggered by sudden stops, hard corners and other risky driving, and the study found that after implementing AlertMeter testing, overall camera activation triggers dropped 17%.

The app startups are competing against a number of companies that have developed other methods for detecting impairment, including goggles and cameras that track eye movement. App makers believe they have an advantage because the ocular devices require expensive hardware. Interest in new impairment-detection technologies has also grown as employers grapple with how to fairly and accurately test for marijuana use in the dozens of states where the drug has recently become legal, said Nina French, a member of Impairment Science’s advisory board.

Testing for marijuana has proven especially challenging, because the drug can trigger a positive test when a user is no longer intoxicated days or weeks after smoking or ingesting it. “Cannabis legalization is expanding, but there are no established methods for detecting cannabis impairment,” Johns Hopkins researchers wrote in a 2021 study using the Druid app. Researchers wrote that the Druid app was the “most sensitive test to cannabis impairment,” compared with classic field sobriety tests and a series of other computer evaluations that required users to replicate shapes using their keyboard, among other tasks. “Standard approaches for identifying impairment due to cannabis exposure ... have severe limitations,” they wrote.

“There is a need to identify novel biomarkers of cannabis exposure and/or behavioral tests like the Druid that can reliably and accurately detect cannabis impairment at the roadside and in the workplace.”

The Druid app instructs users how to take the first step of the test

Impairment Science’s one-minute Druid test paces users through a series of touch-screen tasks: clicking small squares when they blink on the screen or touching an oval shape when they see a circle flash on the screen. Then it reverses the instructions so that a user must tap the oval when they see a square and tap the circles when they blink on screen, which research shows becomes increasingly difficult after several rounds of beer or hits from a joint. It then asks users to tap flashing circles while counting to 15 seconds, and it finishes with instructions to balance on each leg. The phone tracks the users’ accuracy and speed tapping circles and squares, and it measures their steadiness while balancing, similar to a field sobriety test.

Druid calculates a user’s baseline score from the first three times they use the app. From then on, any score dramatically worse than their baseline indicates that they are impaired. The app doesn’t identify why the user is impaired. The test taker’s reduced cognitive ability and depleted motor skills could be due to any number of causes – sleep deprivation, alcohol, marijuana or other drug use. Employers can use the results to prompt workplace policies that may include drug tests, or removal from the road, for example. Organizations can also use the results for business-wide performance analysis.

On the left, the Druid app shows users why they performed better or worse

than usual. On the right, the app shows how a user’s score compares with

average scores and the correlated blood alcohol level.

For example, two fire departments recently used the Druid app to study its current shift and sleep schedules to reduce fatigue among their firefighters. Quick results from apps show promise where Breathalyzer-style devices for detecting recent marijuana use have proven elusive. The once-promising startup Hound Labs launched a test-and-send breath device for detecting recent marijuana use, but it required the sample to be sent to a lab for analysis. The company ran out of money and closed its doors last June before it could launch an on-demand device that would give instant results.

“Unfortunately, with a 10-year history and over $150 million invested, the company required both a significant revenue run rate and delivery of its long-promised OnDemand solution to complete a required funding round,” former Hound Labs CEO Steve Lewis wrote on LinkedIn. “And in a typical early stage Catch 22, without addition-al financing the company couldn’t survive long enough to complete the OnDemand technology which would deliver that required run rate.” French, now a member of Impairment Science’s advisory board, was a former president of Hound Labs. She left the company in 2023 and believes impairment testing apps will be the most likely solution for trucking companies and other employers who need to know if workers are stoned on the job.

French is also the current CEO of startup Neopharma, whose technology uses the phone’s camera to read the indicator strips on drug tests that turn a color when a substance is detected. The app is intended to remove any interpretation bias on the part of a workplace testing administrator, she said. Neopharma’s app integrates with the Druid app to store drug test results with the impairment results in a singular record.

Founded by Dr. Michael Milburn, a retired University of Massachusetts professor who introduced the first version of the Druid app in 2018, Impairment Science gained mainstream attention from a Wired magazine story, but it floundered in attempts to sell the technology to consumers directly, Schiller said. When Schiller became CEO in 2020, he pivoted the company to a business-to-business sales strategy. “Employers have a forcing function on their employees,” he said. “So the company got started for real during Covid – to the extent anything could get started during Covid – and we’ve really only gotten some traction over the last three years.”

Over time, a worker who is required to use Druid daily before getting behind the wheel may slowly, albeit slightly, improve their baseline as they get better standing on one foot, or more comfortable with the flashing circles test. But once inebriated, their score will still suffer compared to their average sober result, he said.

“There’s a limit to how much better they can get,” he said. “There’s diminishing returns, but the important point to make is, once you get a baseline, it is pretty solid, and that makes it a good comparator against your current score. So it’s remarkable how steady people’s baseline is.”

Schiller said the company is doing research now to detect why a user is impaired – a drunk with a poor score on the Druid app likely performs differently than someone who is high and has a poor score, for example.

“When you get two parties, one party is a bunch of drunks and the other is a bunch of stoners, you experience them on the surface as vastly different,” he said. “You’ve got peace and love in one room, and you have a riot in the next room.”