Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Unsupervised continuous sleep analysis.

G Grube1, A Flexer, G Dorffner

  • 1Austrian Research Institute for Artificial Intelligence, Neural Computation Group, Vienna, Austria. georgg@oefai.at

Methods and Findings in Experimental and Clinical Pharmacology
|February 11, 2003
PubMed
Summary

This study introduces a novel, automatic sleep analysis tool using Hidden Markov Models (HMMs) to describe the human sleep-wake continuum. The tool provides a continuous, second-by-second sleep analysis, identifying wakefulness, deep sleep, and REM sleep without human scoring.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Developing a Patient-Centered Model of Prostate Cancer Care: Patient Satisfaction With a Survivorship Program Embedded in Urologic-Oncologic Care.

Urology·2021
Same author

A prediction tool for initial out-of-hospital cardiac arrest survivors.

Resuscitation·2014
Same author

Orexin receptor antagonism, a new sleep-enabling paradigm: a proof-of-concept clinical trial.

Clinical pharmacology and therapeutics·2012
Same author

Risk-neutral density extraction from option prices: improved pricing with mixture density networks.

IEEE transactions on neural networks·2008
Same author

Financial volatility trading using recurrent neural networks.

IEEE transactions on neural networks·2008
Same author

[Technical feasibility of the implantation of a monorail stent system into the renal arteries without pre-dilatation].

RoFo : Fortschritte auf dem Gebiete der Rontgenstrahlen und der Nuklearmedizin·2005

Area of Science:

  • Neuroscience
  • Computational Biology
  • Sleep Medicine

Background:

  • Traditional sleep scoring relies on subjective rules and manual analysis, limiting temporal resolution and objectivity.
  • The EU-funded SIESTA project aimed to develop an automated, high-resolution method for sleep analysis as an alternative to conventional scoring.

Purpose of the Study:

  • To develop and evaluate a fully automatic, probabilistic sleep analyzer using Hidden Markov Models (HMMs) based on single-channel EEG data.
  • To model the human sleep-wake continuum with high temporal resolution, independent of subjective scoring rules.

Main Methods:

  • Utilized a three-state Gaussian Observation Hidden Markov Model (GOHMM) to analyze non-stationary electroencephalogram (EEG) time series.
  • Trained HMMs on data from two distinct sleep laboratories separately to assess laboratory-specific effects.

Related Experiment Videos

  • Constructed pseudo Rechtschaffen and Kales (R&K) hypnograms from model probabilities for comparison with expert scoring.

    • Main Results:

    • The HMM-based analyzer achieved approximately 80% accuracy in detecting key sleep stages (wakefulness, deep sleep, REM sleep) in one laboratory, consistent with previous findings.
    • Training separate models for each laboratory did not improve overall accuracy, highlighting significant laboratory effects likely due to hardware and filter variations.
    • The analyzer provides a continuous, probabilistic, second-by-second quantification of the sleep-wake continuum, capturing major sleep processes without human input.

    Conclusions:

    • The developed HMM approach offers a continuous, data-driven description of human sleep, moving beyond traditional rule-based scoring.
    • Laboratory-specific signal characteristics present a challenge for generalizing automated sleep analysis models.
    • The probabilistic sleep analyzer successfully identifies core sleep states (wakefulness, deep sleep, REM sleep) with high temporal resolution.