1008 Applications of Machine Learning to Improve Time in Bed Detection by Leg-Worn Actigraphy

Ferrée, Thomas C.; Moynihan, Martin; Gozani, Shai N.

doi:10.1093/sleep/zsz067.1005

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An increasing number of wearables and phones promise sleep physiology measurements of some kind with classification accuracy in the range of 67%-92% (Sano et al, 2015). A recent study using machine-learning for actigraphy-based time-in-bed estimation performed at 97.3% classification accuracy (Ferree, Moynihan, & Gozani, 2019) and further improvements in that direction could reduce the need for manual review based on event markers.…”

Section: Discussionmentioning

confidence: 99%

Event marker compliance in actigraphy

Zibrandtsen

Hernández

Ibsen

et al. 2019

Journal of Sleep Research

View full text Add to dashboard Cite

Wrist-worn actigraphy sensors translate measurements of limb acceleration into temporal patterns of wakefulness and sleep. The Actiwatch combines actigraphy with light-sensing capabilities to improve sleep−wake classification (Harrison et al., 2019). Actigraphy is a complementary assessment tool that can help identify changes in sleep patterns specific to certain disorders (Sadeh, 2011), particularly insomnia, hypersomnia and disorders of circadian rhythm. Compared with polysomnography (PSG), binary sleep/wake classification from actigraphy has a sensitivity of 96.5% but only 32.9% specificity, and this is influenced by age, sex, insomnia and sleep/ wake timing (Marino et al., 2013).A U-shaped relationship between health and sleep duration has been suggested (Watson et al., 2015); sleep duration under 6 hr is associated with negative cardio-metabolic outcomes (Grandner et al., 2014). These associations have been challenged due to uncertainty about the causal link between long sleep duration and mortality, and because studies using objective sleep measurement find very few individuals sleeping more than 8.5 hr compared with studies relying on self-report measures (Kurina et al., 2013).Reasons for sleep diary inaccuracy include patients filling out the diary from memory at the end of the monitoring period (Stone et al., 2002). This underscores the importance of objective compared with subjective sleep measures. Actigraphy is valid compared with sleep diaries (Jungquist et al., 2015), and has the added advantage

show abstract

Section: Discussionmentioning

confidence: 99%