About 70 million people report problems with their sleep. And more than ever, they are turning to technology to help them with this plight. Wearable fitness devices are cornering the market, and AI-enabled devices are leading the pack. With flashy terms like machine learning and neural networking backing their products, producers of popular AI sleep trackers are encouraging many consumers to purchase their devices and apps to improve their sleep habits.
We’re spending all this money on seemingly advanced sleep technologies, but are they really helping us sleep better?
Popular technology
According to Adam Blacker, communications lead at Apptopia, almost 11.5 million people (11,490,230, to be exact) downloaded one of the top 10 sleep apps on the Apple iOS platform in the past six months. This astounding number is only a percentage of worldwide sleep tracker use.
The numbers suggest about 37.8 percent of people in the United States are using wearable technology, according to Dr. Jeffrey Durmer, chief medical officer of FusionHealth, a company that aims to help employers address the sleep problems of their workforce. He says, “Generally speaking, sleep tracking technology has helped elevate the importance of sleep for most people, which is essential in the U.S. given the growing public health problem of poor sleep.”
June 5th: The AI Audit in NYC
Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
Promising AI innovations
AI-powered sleep trackers tout the ability to monitor sleep behavior and help consumers change their bad habits. Consumer-facing options like Sleep Watch and Rythm use machine learning models to offer predictive diagnoses for common sleep disorders in consumers. An innovation out of the MIT sleep lab takes this idea a step further, using advanced AI algorithms to track sleep habits without having to hook participants up to a machine or device.
The goal of MIT’s technology is to monitor patient behavior wirelessly and use the results to diagnose sleep disorders. Although MIT’s algorithm is trained to help researchers produce more accurate results than the apps and gadgets available to consumers, even advanced technologies like this might not be accurate enough to diagnose sleep issues.
The problem with AI and sleep tracking
The problem with AI algorithms, according to Michael Larson, founder and president of Sleep Shepherd, is that “the heart of most AI algorithms today is pattern-matching. The problem in the sleep-tech arena is that the patterns used in algorithms are flawed in that they do not characterize sleep well.”
While AI has mastered games, for instance, Larson notes that sleep data is “not as straightforward as the regular patterns which occur in a game.” AI is flawed when it bases its algorithms off of motion sensors, which can give inaccurate data. Therefore, sleep tracker technology needs to come a long way before AI is going to be useful.
Sleep tracker data offers a starting point
If awareness of a problem is the first step in solving it, then AI-enabled sleep trackers are setting us on the path to a solution. Although an algorithm isn’t quite accurate enough to fully diagnose and recommend treatment for sleep issues on its own, the results you get from a tracker could help you become a more informed patient when you consult a physician.
Purchasing a sleep tracker and relying solely on its results to guide your journey to improved sleep is like purchasing a specific mattress based on statistics that suggest it works for others. Just because 40 percent of Americans find a queen size to be the most comfortable doesn’t mean it will be the most comfortable for you. In the same vein, just because a tracker trained on the results of others says you might have a specific sleep problem doesn’t mean it’s true.
You need to go through a lot of trial and error before you can really identify what’s going on with your sleep issues. Results from an algorithm can be just as faulty as relying on statistics to guide your sleep decisions. Sure, they give you a good starting point for assessing your problem, but further research will be necessary to fully understand what’s going on.
Experimentation is key, as well as tracking your overall trends. Dr. Benjamin Smarr, a Reverie sleep researcher, encourages people to go to bed at the same time each night and note how they feel in the morning. Experimenting with caffeine and alcohol intake, bedtime routine, etc., and comparing it to tracker data will give valuable feedback.
A look into the future
Despite potential user error and inaccurate analysis, AI has a bright future in sleep technology. “By learning individual sleep patterns, sleep trackers could potentially utilize AI to better predict sleep/wake states, as well as sleep disturbances such as restless legs syndrome and obstructive sleep apnea,” Durmer says. With the ability to utilize data as a “preclinical” detection device, AI could dramatically improve self-analysis and the detection of those at risk for sleep disorders, he adds.
The potential for connectivity is also limitless. “Not only will sleep trackers help ensure the doors are locked at night and coffee brews in the morning, but they will become more active in the sleep process. Trackers will connect with adjustable bed bases, thermostats, audio devices, and even alarm systems to help control and optimize sleep through environmental control while easing communication and awareness throughout the home,” says Miguel Marrero, director of technology development at REM-Fit.
Artificial intelligence has the potential to bring sleep technology into the future. Devices that can attempt to diagnose sleep disorders, monitor vital signs, and tell a person when to go to bed based on sleep data are here. The only roadblock to their efficacy is a misunderstanding from consumers.
Although it seems like a tracker with cool technology behind it can solve all of your sleep problems on its own, you still need to put in the work on your own and take its insights with a grain of salt. Data from an AI-powered sleep tracker might be able to help you become more informed about your sleep patterns, but decoding what that means still cannot be done by machine learning alone.
Hilary Thompson is a health and wellness consultant and journalist who writes for Today, KSL News, and Girls’ Life.
