A research team published findings on salivary biomarkers for sleep deprivation in the Journal of Proteome Research. A predictive machine learning model, trained on metabolic data from saliva samples, correctly identified samples from sleep-deprived individuals 94 percent of the time.
The study enrolled 20 healthy adult males who typically sleep between seven and nine hours per night. Participants completed three sleep conditions in a randomized sequence, with one-week intervals between each protocol. The protocols included one night without sleep, four nights of two-hour sleep reduction, and a control phase of approximately eight hours of sleep. Researchers collected and analyzed saliva samples for metabolic composition before and after each sleep condition.
Statistical analysis identified 10 distinct molecular differences between saliva samples from sleep-deprived and well-rested participants. The sleep-restricted condition, however, showed no measurable metabolic differences compared to the well-rested state.
The physiological impacts of sleep deprivation can mirror the effects of severe intoxication, and sleep loss reduces human alertness and motor coordination. Fatigued driving contributes to tens of thousands of motor vehicle crashes annually in the U.S. Several U.S. states have enacted legislation aimed at preventing fatigued driving.
Some participants in the study did not return to a baseline metabolic profile after eight hours of recovery sleep following a 24-hour period without sleep. Classification errors in the predictive model were attributed to individual metabolic variation among study participants.
Thomas Kraemer, corresponding author of the study, said, "Until now, sleep deprivation has been impossible to measure biochemically - and yet it is one of the greatest burdens of our time. This study introduces the first direct biomarkers of sleep loss in saliva under real-world conditions, marking a milestone in forensic investigations." The research team plans to evaluate the predictive model using over 1,000 saliva samples collected from shift workers, women, and frequent drivers.
No independent assessment of Thomas Kraemer’s claims was available.