Publication detail

Identification Of Sleep/Wake Stages In Actigraphy Data Utilising Gradient Boosting Algorithm

MIKULEC, M.

Original Title

Identification Of Sleep/Wake Stages In Actigraphy Data Utilising Gradient Boosting Algorithm

Type

conference paper

Language

English

Original Abstract

As the popularity of decentralised clinical trials in-creases, there is a need to have a tool enabling remote assessmentof sleep, while having good consistency with the golden standard, i.e. with polysomnography (PSG). This study aims to introducea new approach to sleep assessment that utilises the modelling ofactigraphy data by a gradient boosting algorithm. The methodis compared to a conventional baseline technique in terms ofsleep/wake stages detection accuracy in a dataset containing 55recordings of actigraphy and PSG (acquired from 28 subjects). Inaddition, we explored how well the outputs of the new methodagree with data acquired via sleep diaries in another datasetincluding 150 recordings (22 subjects). With 97% sensitivity and73%specificity, the new method significantly outperformed thebaseline one in modelling the PSG ground truth. On the otherhand, it had a lower agreement with the patient-reported out-comes. The results suggest that a combination of both approachescould be a good alternative to the golden standard in remote sleepassessment studies.

Keywords

actigraphy; machine learning; polysomnography; sleep; sleep diary

Authors

MIKULEC, M.

Released

27. 4. 2021

Publisher

Brno University of Technology, Faculty of Electrical Engineering and Communication

Location

Brno

ISBN

978-80-214-5943-4

Book

Proceedings II of the 27st Conference STUDENT EEICT 2021 selected papers

Edition

1

Pages from

270

Pages to

274

Pages count

5

URL

https://www.fekt.vut.cz/conf/EEICT/archiv/sborniky/EEICT_2021_sbornik_2.pdf

BibTex

@inproceedings{BUT172033,
  author="Marek {Mikulec}",
  title="Identification Of Sleep/Wake Stages In Actigraphy Data Utilising Gradient Boosting Algorithm",
  booktitle="Proceedings II of the 27st Conference STUDENT EEICT 2021 selected papers",
  year="2021",
  series="1",
  pages="270--274",
  publisher="Brno University of Technology, Faculty of Electrical Engineering and Communication",
  address="Brno",
  doi="10.13164/eeict.2021.270",
  isbn="978-80-214-5943-4",
  url="https://www.fekt.vut.cz/conf/EEICT/archiv/sborniky/EEICT_2021_sbornik_2.pdf"
}