Scientists from the Sleep Research Centre at the University of Surrey say they used a machine learning algorithm to identify a subset of 68 genes to detect with 92% accuracy whether a blood sample was from a sleep-deprived or well-rested individual.
The study (“Identifying and validating blood mRNA biomarkers for acute and chronic insufficient sleep in humans: a machine learning approach”) appears in the journal Sleep.
“Acute and chronic insufficient sleep are associated with adverse health outcomes and risk of accidents. There is, therefore, a need for biomarkers to monitor sleep debt status. None are currently available. We applied Elastic-net and Ridge regression to entire and pre-filtered transcriptome samples collected in healthy young adults during acute total sleep deprivation and following one week of either chronic insufficient (< 6 h) or sufficient sleep (~8.6 h) to identify panels of mRNA biomarkers of sleep debt status. The size of identified panels ranged from 9–74 biomarkers. Panel performance, assessed by leave-one-subject-out cross-validation and independent validation, varied between sleep debt conditions. Using between-subject assessments based on one blood sample, the accuracy of classifying ‘acute sleep loss’ was 92%, but only 57% for classifying ‘chronic sleep insufficiency’. A reasonable accuracy for classifying ‘chronic sleep insufficiency’ could only be achieved by a within-subject comparison of blood samples. Biomarkers for sleep debt status showed little overlap with previously identified biomarkers for circadian phase,” write the investigators.
“Biomarkers for acute and chronic sleep loss also showed little overlap but were associated with common functions related to the cellular stress response, such as heat shock protein activity, the unfolded protein response, protein ubiquitination and endoplasmic reticulum associated protein degradation, and apoptosis. This characteristic response of whole blood to sleep loss can further aid our understanding of how sleep insufficiencies negatively affect health. Further development of these novel biomarkers for research and clinical practice requires validation in other protocols and age groups.”
The discovery paves the way for a future test which will be able to assess if a driver was sleep deprived, according to the team. Previous research in this area from the AAA Foundation for Traffic Safety has shown that drivers who get just one to two hours less than the recommended daily allowance in a 24-hour period nearly double their risk for a car crash.
Emma Laing, Ph.D., senior lecturer in bioinformatics at the University of Surrey, says “We all know that insufficient sleep poses a significant risk to our physical and mental health, particularly over a period of time. However, it is difficult to independently assess how much sleep a person has had, making it difficult for the police to know if drivers were fit to drive, or for employers to know if staff are fit for work.”
“Identifying these biomarkers is the first step to developing a test which can accurately calculate how much sleep an individual has had,” adds Simon Archer, Ph.D., professor of molecular biology of sleep at the University of Surrey. “The very existence of such biomarkers in the blood after only a period of 24-hour wakefulness shows the physiological impact a lack of sleep can have on our body.”
“This is a test for acute total sleep loss,” notes Derk-Jan Dijk, Ph.D., director of the Sleep Research Centre. “The next step is to identify biomarkers for chronic insufficient sleep, which we know to be associated with adverse health outcomes.”
