D4 Published development or research report or study
Validation of an Accelerometer Based BCG Method for Sleep Analysis
Authors: Sami Nurmi, Tarja Saaresranta, Tero Koivisto, Ulf Meriheinä, Lauri Palva
Publisher: Aalto University
Publication year: 2016
Series title: Science + Technology
Number in series: 7
First page : 1
Last page: 12
eISBN: 978-952-60-6842-8
ISSN: 1799-4896
Web address : http://urn.fi/URN:ISBN:978-952-60-6842-8
Self-archived copy’s web address: http://urn.fi/URN:ISBN:978-952-60-6842-8
Sleep problems are one of the most common medical complaints today.
Polysomnography (PSG) as the current standard for sleep analysis is
expensive, intrusive and complex. Thus, finding a reliable and
unobtrusive method for longer-term home use is important.
Ballistocardiography (BCG) based methods have shown potential in sleep
analysis recently. The usability and performance of a BCG based method
in qualitative and quantitative analysis of sleep was evaluated. The
method was validated in a clinical test on 20 subjects using PSG as a
reference. Heart rate (HR), heart rate variability (HRV), respiratory
rate (RR), respiratory rate variability (RRV), respiratory depth
(Rdepth) and movement were utilized for sleep stage detection.
BCG parameter accuracy was presented as the mean error from PSG with
95% confidence interval. The errors were -0.1 ± 4.4 beats per minute for
HR, -0.9 ± 14.7 ms for high frequency (HF) HRV, -3.0 ± 29.9 ms for low
frequency (LF) HRV, 0.3 ± 4.5 breaths per minute for RR and -40 ± 424 ms
for RRV respectively. Correlation coefficient was 0.97 for HR, 0.67 for
HF HRV, 0.71 for LF HRV, 0.54 for RR and 0.49 for RRV. HR, RRV and
Rdepth were typically at an increased level in REM sleep and wakefulness
and decreased in deep sleep. RRV was at its highest during wakefulness.
HRV was at a decreased level in REM and wakefulness and increased in
deep sleep. Movement was higher during wakefulness than in sleep.
