Kleine-Levin syndrome: warning signs found 48 h ahead

Salzburg, October 2026. Kleine-Levin syndrome is one of the rarest sleep disorders, and one that a sleep laboratory can hardly capture. A 22-year-old patient has now slept around 100 nights with a sensor on his upper arm and the sleep-staging analysis of sleep². The result: 20 attack nights (nights with more than 15 hours of sleep) were recorded, and in the 48 hours before an attack a pattern emerged in deep sleep and heart rate that the patient himself does not notice. Univ.-Prof. Dr. Manuel Schabus, sleep researcher at the University of Salzburg and co-founder of sleep², uses the case as an example in an overview article on AI and wearables in sleep medicine for the physicians’ journal psychopraxis.neuropraxis (Springer, in press). For him it is the most striking example of what daily measurement can achieve.

What Makes Kleine-Levin Syndrome So Elusive

Kleine-Levin Syndrome (KLS), often called the "Sleeping Beauty Syndrome" in the media, is a rare central hypersomnia that usually begins in adolescence. During episodes, affected individuals sleep 12 to 24 hours a day, accompanied by cognitive and behavioral changes. Between episodes, there are weeks to months of complete inconspicuousness.

This exact pattern makes diagnosis so difficult. The diagnosis is made clinically; polysomnography and sleep latency tests are usually normal between episodes and inconsistent during episodes. Biomarkers for onset, course, or recovery are lacking. Or, as Manuel Schabus puts it in his article: One would have to be in the lab when the episode comes – and no one knows when it will come.

100 Nights, 20 Attack Nights: What Long-Term Measurement Revealed

The patient wore an optical sensor on his upper arm (Polar Verity Sense) for about 100 nights. Sleep stages were calculated from heart rate intervals – using the method that the Salzburg research group trained on 8,898 manually evaluated lab nights (Topalidis et al., 2023, Sensors) and that achieves about 84 percent agreement with polysomnography at home across four sleep stages (Topalidis et al., 2025, Preprint). We have described elsewhere how the four sleep phases are reflected in heart rhythm and what sleep trackers can reliably do today.

During this period, 20 attack nights occurred: nights with more than 15 hours of sleep that clearly stood out from the baseline nights in the course chart (Kunz & Topalidis, neurological 3/2025). The complete scientific case report (Schabus et al.) has been submitted for publication and can soon be read in detail. Making it visible is already a diagnostic gain. However, the actual step begins thereafter.

48 Hours in Advance: A Pattern the Patient Does Not Feel

Those who have 100 nights of sleep architecture, heart rate, and heart rate variability can ask what precedes an attack – without having to know in advance where to look. This is where the strength of systematic course analysis lies: It compares all nights before an attack with all other nights, finding patterns that remain invisible in a single night, both in the lab and at home, and that the patient himself does not feel.

In an exploratory analysis, the clearest signal appeared about 48 hours before the attacks: the proportion of deep sleep increased from 11.1 to 15.1 percent (p = 0.002, d = 0.86), and the nightly heart rate shifted upwards: the proportion of sleep time with a heart rate over 80 beats per minute increased from 53.5 to 63.2 percent (p = 0.053, d = 0.58), statistically just on the verge of significance, but with a medium effect size. A prodromal signature that not even the patient perceives.

Important for classification: This is a single-case analysis, exploratory and hypothesis-generating. The sleep stages are algorithmically estimated and not confirmed by EEG. Whether the pattern also occurs in other affected individuals must be shown by further cases and prospective studies. The Salzburg research group is therefore looking for more KLS patients willing to measure their sleep at home over a longer period to test the pattern in additional patients. Contact: hello@sleep2.com.

Why This Case Points Beyond Kleine-Levin Syndrome

"In rare, episodic diseases, for which group studies are hardly possible, dense course measurement in the individual can become the actual data source," writes Manuel Schabus. Early warning systems that alert affected individuals and relatives before an episode, more precise phenotyping through course patterns, and prospective studies with multiple patients are conceivable.

The same principle can be transferred: to narcolepsy with its daytime sleep attacks, to night-to-night variable sleep apnea, to circadian disorders, to parasomnias like sleepwalking, whose episodes almost never occur in the lab night – and to insomnia itself, whose definition requires disturbed sleep on at least three nights per week, a pattern that a single lab night only hits by chance.

Polysomnography remains the gold standard for diagnosing sleep-related breathing and movement disorders and parasomnias. "Wearables and artificial intelligence will not replace sleep medicine – but they expand it by a dimension that has been missing so far: time," says Schabus.

Dr. Thomas Winkler, CEO of sleep²: "This case shows where objective sleep measurement is developing: from the snapshot to the course. We want to make this possibility accessible to clinics, research groups, and partners in the field of rare diseases."