Research
Print page Print page
Switch language
The Capital Region of Denmark - a part of Copenhagen University Hospital
E-pub ahead of print

Non-invasive machine learning estimation of effort differentiates sleep-disordered breathing pathology

Research output: Contribution to journalJournal articleResearchpeer-review

DOI

  1. Multi-frequency bioimpedance in equine muscle assessment

    Research output: Contribution to journalJournal articleResearchpeer-review

  2. A maximum entropy method to compute the 13NH3 pulmonary transit time from right to left ventricle in cardiac PET studies

    Research output: Contribution to journalJournal articleResearchpeer-review

  1. Sleep in cluster headache revisited: Results from a controlled actigraphic study

    Research output: Contribution to journalJournal articleResearchpeer-review

  2. Assessment of diurnal melatonin, cortisol, activity, and sleep-wake cycle in patients with and without diabetic retinopathy

    Research output: Contribution to journalJournal articleResearchpeer-review

  3. Disturbed sleep in cluster headache is not the result of transient processes associated with the cluster period

    Research output: Contribution to journalJournal articleResearchpeer-review

  • Umaer Hanif
  • Logan Douglas Schneider
  • Lotte Trap
  • Eileen B Leary
  • Hyatt Moore Iv
  • Christian Guilleminault
  • Poul Jørgen Jennum
  • Helge Bjarup Dissing Sørensen
  • Emmanuel J M Mignot
View graph of relations

OBJECTIVE: Obstructive sleep-disordered breathing (SDB) events, unlike central events, are associated with increased respiratory effort. Esophageal pressure (Pes) monitoring is the gold standard for measuring respiratory effort, but it is typically poorly tolerated because of its invasive nature. The objective was to investigate whether machine learning can be applied to routinely collected non-invasive, polysomnography (PSG) measures to accurately model peak negative Pes.
 Approach: 1119 patients from the Stanford Sleep Clinic with PSGs containing Pes served as the sample. The selected non-invasive PSG signals included nasal pressure, oral airflow, thoracoabdominal effort, and snoring. A long short-term memory (LSTM) neural network was implemented to achieve a context-based mapping between the non-invasive features and the Pes values. A hold-out dataset served as a prospective validation of the algorithm without needing to undertake a costly new study with the impractically invasive Pes. 
 Main results: The median difference between the measured and predicted Pes was 0.61 cmH2O with an interquartile range (IQR) of 2.99 cmH2O and 5th and 95th percentiles of -5.85 cmH2O and 5.47 cmH2O, respectively. The model performed well when compared to actual esophageal pressure signal (ρmedian=0.581, p=0.01; IQR = 0.298; ρ5% = 0.106; ρ95% = 0.843).
 Significance: A significant difference in predicted Pes was shown between normal breathing and all obstructive SDB events; whereas, central apneas did not significantly differ from normal breathing. The developed system may be used as a tool for quantifying respiratory effort from the existing clinical practice of PSG without the need for Pes, improving characterization of SDB events as obstructive or not.&#13.

Original languageEnglish
JournalPhysiological Measurement
ISSN0967-3334
DOIs
Publication statusE-pub ahead of print - 2019

ID: 56596801