A Smartwatch App for Monitoring Intoxication
Asahi Breweries and UC Irvine are collaborating on research into a smartwatch-based application for detecting alcohol levels using hyperdimensional computing.
What if your smartwatch could send an alert when you’ve had too much to drink, warning you not to get behind the wheel of a car and drive? This is exactly the type of application researchers from UC Irvine’s Donald Bren School of Information and Computer Sciences (ICS) are working to develop in collaboration with Asahi Brewers, an international food and beverage company based in Tokyo.
“Asahi was looking to use wearables, in particular watches, to verify when someone may be under the influence,” says Computer Science Professor Sergio Gago-Masague. The company’s goal was to encourage healthier habits and increase people’s awareness of their drinking behaviors.
The collaboration led to a study of ways to predict blood alcohol concentration (BAC) using sensors available in a commercial smartwatch, such as an Apple Watch. “Other non-intrusive monitoring applications may require additional sensors, such as transdermal sensors that measure dermal alcohol concentration. DAC can measure the alcohol you metabolize by your skin, but that usually takes 30 minutes,” explains Gago-Masague. “So what’s novel about our approach is its ‘just in time’ intervention that leverages predictive algorithms using non-direct measures, such as motion, location, heart and respiration rates.”
Their application, which uses hyperdimensional computing to predict if a user is under the influence of alcohol, has achieved 93.5% accuracy on average. The researchers have outlined their findings in two recent papers: “Smartwatch-Based Prediction of Transdermal Alcohol Levels Using Hyperdimensional Computing,” presented at the 2024 IEEE 10th World Forum on Internet of Things, and “Enhanced Detection of Transdermal Alcohol Levels Using Hyperdimensional Computing on Embedded Devices” presented at the 2024 International Joint Conference on Neural Networks.
“We are leveraging models in machine learning that can run on smartphones without the need of a cloud service,” says Gago-Masague, referring to his team’s expertise in not only hyperdimensional computing but also machine learning, embedded systems, and security and privacy in health monitoring. The team is now working with UC Irvine’s Beall Applied Innovation to patent the application and model, and they plan to publish an iOS application in the App Store so more people can start using it. “There is so much potential in terms of both alerting users when it’s not safe to drive as well as helping them monitor their health and signs of alcohol abuse.”
Gago-Masague is excited to see the application’s potential realized. “It’s very promising work, and we look forward to continuing our collaboration with Asahi,” he says. “We’ve seen a positive impact on our research into machine learning and edge computing, as well as the potential for a positive impact on society with a business model to go with it. It’s a win-win international collaboration.”
— Shani Murray