Sleep estimation from wrist movement quantified by different actigraphic modalities
Introduction
The emergence of diverse wrist-activity monitors, each with its unique features and modality of operation, has given rise to several methodologic difficulties in actigraphic research. One problem concerns direct comparisons of actigraphic findings when data were gathered with different activity monitors or when differing scoring algorithms were used to handle the data. Important differences among monitors include modality of movement quantification, sampling frequency, sensitivity and filter settings, motion sensor, and directional sensitivity (Tryon, 1991, Sadeh et al., 1995, Teicher, 1995, Van Someren et al., 1996, Pollak et al., 1998).
Generally, activity monitors quantify movement with one of three algorithms: zero crossing, time above threshold, or digital integration. Simultaneous quantification of movement in several modes is also possible with new models of Mini Motionloggers. That these modalities can be used interchangeably in quantifying the vigor of wrist movement or in monitoring sleep and wakefulness has not been demonstrated. Indeed, a recent simulation study found that these three quantification modalities generated dissimilar activity counts in response to the same pattern of activity (Gorney and Allen, 1999). Also important was the observation that digital integration provided a more accurate representation of the intensity of the measured activity (Gorney and Allen, 1999).
Whether such differences would be observed were natural wrist movements analyzed remains unclear. It is equally unclear whether transformations of the transduced activity signal by these quantification modalities affects the accuracy of sleep detection. The present investigation compared sleep–wake estimates from nocturnal wrist movement quantified by different actigraphic modalities. Actigraphic sleep estimates were referenced to contemporaneous polysomnographic (PSG) sleep estimates.
Section snippets
Participants and procedures
Five volunteers (mean age=25, SD=6) without active medical conditions, emotional illnesses, or sleep disturbances participated in the study. Volunteers spent 5 nights and 4 days in the Circadian Pacemaker Laboratory at UCSD, as part of an on-going study. EEG was recorded during in-bed periods (∼8 h) matched to each volunteer's in-bed times in the previous week. Volunteers provided written informed consent approved by UCSD's Institutional Review Board and were paid for their participation.
PSG recording
During
Results
Polysomnographic total sleep time averaged 451 min and sleep efficiency averaged 94.8%. As shown in Table 2, overall the scoring formula for each quantification modality performed accurately in detecting sleep and wakefulness from wrist activity. Estimates from MAXACT-quantified activity showed the lowest agreement rate, and PSG sleep percentages were underestimated. Sleep percentages were slightly overestimated using activity quantified by SUMACT, by ZCM, and by TAT. No bias in detecting sleep
Discussion
This study investigated the validity of sleep–wake estimates obtained from five movement quantification modalities using two commercially available actigraphs. Notwithstanding the necessity for optimizing algorithms, both devices yielded accurate and valid activity data relative to polysomnographic recordings. This suggests that the transducers of these two devices (i.e. single-axis, with or without sensitivity in other measurement axes) performed comparably for monitoring nocturnal sleep and
Acknowledgements
This research was supported by AG12364, HL55983, AG15763, and HL61280. We thank Geralyn Wallace-Guy, Tony Cress, and Robert Fell for their assistance in the study. Ambulatory Monitoring, Inc. donated equipment.
References (21)
- et al.
Gravitational artefact in frequency spectra of movement acceleration: implications for actigraphy in young and elderly subjects
J. Neurosci. Methods
(1996) - et al.
A comparison of sleep detection by wrist actigraphy, behavioral response, and polysomnography
Sleep
(1997) - et al.
Automatic sleep/wake identification from wrist activity
Sleep
(1992) - et al.
What is activity count?: a comparison of different methodologies used in wrist actigraphy
Sleep
(1999) - et al.
Sleep–wake estimation: actillume vs polysomnography
Sleep
(1999) - Jean-Louis G, Kripke DF, Cole RJ, Assmus JD, Langer RD. Sleep detection with an accelerometer actigraph: comparisons...
- et al.
Determination of sleep and wakefulness with the actigraph data analysis software (ADAS)
Sleep
(1996) - et al.
Prevalence of sleep disordered breathing in ages 40–64 years: a population-based survey
Sleep
(1997) - et al.
Comparison of the Actillume and EEG for identifying total sleep time and wake after sleep onset
Sleep Res.
(1995) - et al.
Direct comparison of two widely used activity recorders
Sleep
(1998)