The researchers have an interesting hypothesis. Is mental illness manifested in how we sleep? Can we detect mental illness by closely studying the electrical activity between the brain’s neurons while a person is asleep? And can this data be used for diagnostics?

These are questions that a group of researchers from Aarhus University are trying to answer as part of a new project involving patients with bipolar disorder who are being treated at the psychiatric department at Aarhus University Hospital.

The researches intend to use a small earbud to measure EEG signals in the brain while patients are asleep.

"We need to study how the sleep patterns of patients develop over the course of long-term treatment, where the differences from night to night can vary greatly. Our long-term goal is to provide doctors with an objective indicator of how a disease is developing, so they have a tool for more precise treatment and possibly also more accurate diagnostics," says Kaare Mikkelsen, assistant professor at the Department of Electrical and Computer Engineering at Aarhus University.

He is part of a research group at Aarhus University that has spent several years developing technology to measure EEG signals via the ear and process large amounts of data from the brain over an extended period of time.

This technology is now being taken into use for the first time in a psychiatric context. The project will track 20 patients for six months.

The project is being funded by the Lundbeck Foundation.

Artificial intelligence will help identify sleep signatures in patients

Doctors already know that mental illness and sleep issues often go hand in hand. But they do not always know which comes first: nor do they know exactly how sleep affects mental function. Or vice versa.

Previous studies have shown a correlation between disruptions to our circadian rhythm and the risk of developing mental illness. However, those studies were typically based on self-reported activity and sleep and not data from the brain itself.

Bipolar patients often have very irregular circadian rhythms, and they are therefore an interesting target group for the researchers, who are about to begin collecting the hitherto largest amount of sleep data within psychiatry. Once the data has been collected, it will be converted into detailed sleep reports, also known as hypnograms, which can show how the different sleep stages change over time in the patients.

"It's incredibly exciting because we’ll have access to a lot of data from a specific patient group, which means we should be able to identify patterns. We’re hoping that artificial intelligence can help us identify a sleep signature for bipolar disorder," says Kaare Mikkelsen.

It will be a major breakthrough if the researchers can identify a sleep signature for bipolar disorder, because it is very likely they will then also be able to identify sleep signatures for other broad categories of mental disorders and conditions such as ADHD, schizophrenia, autism and dementia.

"EEG measurements could become an important objective diagnostic tool, and doctors will be able to measure the effect of their medical and psychological interventions more precisely and perhaps even detect relapses," says Kaare Mikkelsen.

A technology race

Kaare Mikkelsen and his research group are not alone in the hunt for new knowledge about our sleep. Something resembling a global technology race has emerged over the past few years. The researchers who discover the most effective and accurate way of measuring what happens in our brains when we sleep will have access to a goldmine of new knowledge about sleep and how it affects our physiological and psychological functions.

Some researchers have tried using headbands. Others have tried to intercept signals from the brain by placing stickers fitted with electrodes to a person’s face. However, so far it looks as if the group of Danish researchers and their earplugs have cracked the code on how to measure EEG precisely and without disturbing sleep. This was demonstrated by a study of 80 full-night recordings of sleep in healthy people, where researchers compared the data quality from the ear EEG with traditional EEG measurements on the scalp.

"Our earplugs measure the differences in electric potential in the brain and we have harvested some very reliable results. We already have lots of anonymised and high-quality sleep data from various tests, and our work on developing the algorithms that will translate the electrical signals into hypnograms has come a long way," says Kaare Mikkelsen.

"Have you slept well?" 

So today we have access to more than five year's sleep data, but it actually gives rise to more questions than answers and perhaps even a showdown with the clinical defintions of sleep quality we use today. 

Because hhat is good sleep? When the researchers ask test subjects whether they have slept well, there is no clear correlation between their subjective experience and the results of the hypnograms.

"I believe this technology will help us create new definitions of sleep quality based on data and evidence. And I think that once we get a deeper and more accurate understanding of sleep and its importance, we’ll be able to help patients and others get a better night’s sleep,” he says.