The Mignot Nature Communications (MNC) dataset contains raw polysomnography data from an automated sleep staging project using neural networks. The piece was published in Nature Communications and the abstract is below.
Analysis of sleep for the diagnosis of sleep disorders such as Type-1 Narcolepsy (T1N) currently requires visual inspection of polysomnography records by trained scoring technicians. Here, we used neural networks in approximately 3,000 normal and abnormal sleep recordings to automate sleep stage scoring, producing a hypnodensity graph—a probability distribution conveying more information than classical hypnograms. Accuracy of sleep stage scoring was validated in 70 subjects assessed by six scorers. The best model performed better than any individual scorer (87% versus consensus). It also reliably scores sleep down to 5 s instead of 30 s scoring epochs. A T1N marker based on unusual sleep stage overlaps achieved a specificity of 96% and a sensitivity of 91%, validated in independent datasets. Addition of HLA-DQB1*06:02 typing increased specificity to 99%. Our method can reduce time spent in sleep clinics and automates T1N diagnosis. It also opens the possibility of diagnosing T1N using home sleep studies.
The National Sleep Research Resource is grateful to Emmanual Mignot for sharing these data.
When using this dataset, please cite the following:
Please include the following text in the Acknowledgements:
The Mignot Nature Communications research was mostly supported by a grant from Jazz Pharmaceuticals to E.M. Additional funding came from: NIH grant R01HL62252 (to P.E.P.); Ministry of Science and Technology 2015CB856405 and National Foundation of Science of China 81420108002,81670087 (to F.H.); H. Lundbeck A/S, Lundbeck Foundation, Technical University of Denmark and Center for Healthy Aging, University of Copenhagen (to P.J. and H.B.D.S). Additional support was provided by the Klarman Family, Otto Mønsted, Stibo, Vera & Carl Johan Michaelsens, Knud Højgaards, Reinholdt W. Jorck and Hustrus and Augustinus Foundations (to A.N.O.). The National Sleep Research Resource was supported by the National Heart, Lung, and Blood Institute (R24 HL114473, 75N92019R002).
The NSRR team harmonized the publicly available EDF and staging data using the Luna software package to make future analyses simpler. The montage and sampling rate information page details the harmonized EDF channel labels. We converted manual sleep staging data into two different formats: .eannot and .xml. Some cohorts did not include manual sleep staging data.
To access the NSRR harmonized versions of the MNC data, complete a new data request. Once the NSRR team approves your access, click here to browse and download files.
The cohorts_deid.xlsx file was presented with the manuscript for matching case and control subjects from the various cohorts. It matches individual subject IDs with their corresponding cohort, a narcolepsy or control diagnosis, and which part or parts of the analysis their sleep study was used for (e.g. training, testing, validation).
CSF refers to cerebrospinal fluid, and if there is a numeric value it would be in reference to how much hypocretin was found there. It is used as a final indicator for narcolepsy.
DB0602 is in reference to the allele HLA-DQB1*06:02; which is relatively common in the general population, and extremely common among those diagnosed as narcoleptic. In other words, most people who are DQB0602 positive do not have narcolepsy, while almost all people with narcolepsy are DQB0602 positive.
Demographic data are not available for MNC subjects.
January 22, 2021
cohorts_deid.xlsx
file to fix data errorsPlease reach out to us at support@sleepdata.org or in the Forum if you have questions.