Capturing autism’s sleep problems with devices nearable and wearable

Next-generation trackers could realize a long-standing research dream: conducting sleep studies in large numbers of autistic people.

A young woman sleeps at home wearing a Dreem headband and surrounded by stuffed animals.
Photography by Simon Simard
Listen to this story:

Autism scientists have long been on a quest to crack the condition’s sleep conundrum. Problems with falling and staying asleep are not typically considered a core trait of autism, but they are exceedingly common among autistic people and have compound effects: They can exacerbate a range of autism traits and are linked to greater difficulties with daily functioning.

But to track and assess what goes on in the body during sleep — historically at least — research participants have had to sleep in a lab, tucked up in a web of cumbersome equipment: The gold-standard approach, polysomnography, temporarily turns a sleeper into something resembling the prey of a giant electrical spider, bound in electrodes and wire leads that capture brain activity via electroencephalography (EEG), and breathing and body position, among other measures.

This scenario is uncomfortable enough for neurotypical people, never mind someone with autism who may also have anxiety and sensory, communication or behavioral difficulties, notes Beth Ann Malow, professor of neurology and pediatrics at Vanderbilt University in Nashville, Tennessee.

“The question is, are you really getting a valid night’s sleep [under those conditions]?”

Some study participants with autism may not be able to sleep at all, notes Thomas Frazier, professor of psychology at John Carroll University in University Heights, Ohio, as was the case when Frazier’s teenage autistic son went to a clinic for an overnight EEG study. “Getting good data and then scaling that up so you can actually use it in a real, meaningful way is tough,” he says.

Clinical Research Coordinator Anabella Tolosa wears an EEG helmet in the sleep room of the Manoach Lab.
Wire web: EEG helmets used for standard polysomnography can interfere with a study participant’s sleep, particularly if she has sensory sensitivities or anxiety.

To address the issue, multiple labs are working to deploy minimally invasive sleep-tracking devices — both “wearables” that someone puts on and “nearables” that take measurements from a distance. These devices tend to collect fewer types of data than polysomnography can, but their creators say they do enough to get the job done, tracking sleep stages and certain aspects of a sleeper’s physiology and brain activity.

The three technologies described below all strike that balance between accuracy and scalability, and they may help researchers increase the number of autistic participants in their studies. “Having a core set of measures that you think you can consistently get completed in the real environment, and then allowing you to scale up to larger sample sizes so you can really appreciate and understand that heterogeneity across individuals, is probably going to be more useful than what we’ve been doing the last 20 years,” Frazier says.

“I’m glad that people are actually coming up with real solutions,” he adds. “We’ve been waiting for this for a decade or so.”

Dreem sleep

During the day, our experiences accumulate in our memory, and at night, they are consolidated while we sleep. This consolidation is mediated by sleep spindles generated in the thalamus, slow waves that propagate from the cortex and ripples that radiate from the hippocampus, says Dimitrios Mylonas, instructor in psychology and a researcher in Dara Manoach’s lab at Massachusetts General Hospital in Boston. “You have a discussion between these different structures of the brain.”

Dimitrios Mylonas examines sleep data on a computer monitor.
Wave watch: By targeting miscommunication between the thalamus and hippocampus during sleep, Dimitrios Mylonas aspires to ease core autism traits.

Mylonas and his colleagues in the Manoach lab are eavesdropping on this conversation via the Dreem headband, which contains EEG sensors to track sleep architecture and an accelerometer to track a wearer’s respiration, head motion and head position. Their overarching goal is to spot where there’s miscommunication so they can correct it and maybe even ease core autism traits or improve cognition.

The crosstalk among sleep oscillations is disrupted in the brains of children with autism, according to a 2022 study Manoach and her team published in Sleep last year. By way of traditional polysomnography, they found altered spindles in the autistic children compared with their neurotypical peers, indicating some difference in the circuit that unites the thalamus and cortex. Now they plan to use the Dreem to dig deeper, and they have received funding from the Autism Science Foundation’s new profound autism grant and the Simons Foundation. (The Simons Foundation is Spectrum’s parent organization.)

Because the thalamus’ spiky spindles appeared to be muted in autistic children, the researchers want to tweak these fast oscillations, Mylonas says. “These sleep spindles are amenable to be modulated.”

Dara Manoach discusses results with team members Dimitrios Mylonas and Paige Hickey.
Dreem team: Dara Manoach, center, discusses results with team members Mylonas (left) and Paige Hickey (right).

To start, they plan to use immediate feedback from the Dreem’s EEG readout to conduct the chorus of sleep oscillations by way of closed loop auditory stimulation. Such auditory stimulation accelerated slow oscillations in non-autistic people, and the magnitude of oscillation change correlated with memory improvements in the participants in one 2018 study.

The 2022 Sleep study set them on this path but was limited by a small sample size, Mylonas says. Moving forward, home-based wearables should not only expand the possible number of participants, but also make the results more “ecologically valid,” he says — allowing researchers to observe the sleep oscillations of people sleeping in their own beds.

Night watch

Smartwatches may also help researchers better understand the link between autism and sleep difficulties. The relationship has been widely reported, but its direction — whether poor sleep contributes to any autism-linked traits or vice versa — remains unclear, says Ilan Dinstein, associate professor of psychology at Ben-Gurion University of the Negev in Israel.

Image of a white Embrace Plus smartwatch.
Smartwatch option: The EmbracePlus gives researchers access to the raw data it collects.

Past research has suggested that sleep is important for cognitive abilities. But when he and his colleagues ran a study of the relationship between autism traits and poor sleep in autistic children, they found that sensory sensitivities and irritability were more strongly tied to disturbed sleep than were cognitive ability or core autism traits. “I was really surprised by this,” Dinstein says.

The study assessed children’s sleep habits using parent reports, which are subjective measures, so Dinstein and his colleagues sought a more objective way to evaluate different sleep parameters.

They turned to actigraphy monitors — wearable devices, such as Fitbits, that track a person’s movements. The monitors are small and can be worn on a wrist or ankle, or even taped to another body part, making them ideal for continuously tracking movement and establishing a person’s circadian rhythms, Dinstein says. But standard actigraphy monitors can provide only a rough snapshot of sleep duration and awake periods; for example, they cannot reliably identify when a person first falls asleep, he says. “The limitation is that you could be inactive for multiple reasons, and not necessarily because you’re sleeping.”

Instead they adopted a newer actigraphy monitor that reports additional physiological data to more accurately determine whether a person is awake or asleep: The EmbracePlus smartwatch tracks a person’s movement, skin conductance, body temperature and heart rate, among other metrics, and provides the researchers with access to the device’s raw data.

They also plan to combine the actigraphy with another measure, such as EEG from the Dreem headband, to create an even fuller portrait of a person’s sleep behavior — and more akin to the data obtained in a traditional sleep study. (Dinstein has also received funding from the Simons Foundation for his research.)

Dinstein and his colleagues are piloting this pairing of the EmbracePlus and the Dreem headband on non-autistic adults and later plan to use the devices in tandem to track sleep in autistic adults and children. The eventual goal, Dinstein says, is to run an intervention study to see whether improving autistic people’s sleep eases other autism-linked traits.

“The really interesting question is to what degree the change in sleep” will affect other behavioral domains, he says.

Making waves

Even in short studies on sleep, having participants wear EEG headbands was a challenge, says Dina Katabi, director of the Center for Wireless Networks and Mobile Computing at the Massachusetts Institute of Technology. “Without even knowing about it, they just rip it off their head while they sleep.”

But what if a sleep tracker didn’t even need to touch the body of the person it tracks? The Emerald, which mounts on a wall and resembles a wireless internet router, was born in Katabi’s lab to test that question. The device sends radio waves into the surrounding space and tracks how a person’s movements distort the waves’ return, not unlike how sonar maps what lies below a ship.

Unlike polysomnography, the Emerald does not create a real-time readout of someone’s brain activity, heart rate, motion or breathing. Instead, the device feeds its data through an artificial intelligence processing program trained on polysomnography data. The lab has validated the device in neurotypical people, and it can measure disease severity and medication responses in people with Parkinson’s disease, according to a pair of 2022 papers from Katabi’s lab that used the Emerald to track gait and nighttime breathing. The device can also track sleep and breathing in people with Rett syndrome, Katabi says, but those data are not yet published.

The Emerald sleep tracker, a white box, sits on a blue workbench surrounded by electronic equipment.
Wall mount: The Emerald sleep tracker makes use of radio waves to monitor someone’s slumber.

People with Rett syndrome tend to have trouble getting to sleep and staying asleep, and their breathing can be especially shallow, which leads to hyperventilation. They may also temporarily stop breathing — a phenomenon known as sleep apnea — and the Emerald can detect these changes in breathing, Katabi says. In February, the Rett Syndrome Research Trust gave Emerald Innovations more than $1 million to further develop the device for clinical trials. Hopefully, the device will help researchers discern whether therapies are working in a clinical trial, says Jana von Hehn, chief scientific officer at the Rett Syndrome Research Trust.

The Emerald is ready for use in the trial, Katabi says, and the validation data showed such a big difference between people with and without Rett syndrome that the results were statistically significant even with a relatively small number of participants.

“They can monitor unobtrusively, as long as you pay the electricity bill, so in that sense, that is huge,” says Bernhard Suter, assistant professor of pediatric neurology at Baylor College of Medicine in Houston, Texas. “It would truly capture, in real time, changes in a fairly natural environment.”

With additional reporting by Rebecca Horne.