Validation of an innovative method, based on tilt sensing, for the assessment of activity and body position. (June 2015)

• Record Type: Journal Article
• Authors: Bonmati-Carrion, M. A.; Middleton, B.; Revell, V. L.; Skene, D. J.; Rol, M. A.; Madrid, J. A.
• Is Part Of: Chronobiology international. Volume 32:Number 5(2015)
• Page Start: 701
• Page End: 710
• Publication Date: 2015-06
• Subjects: Chronobiology -- Periodicals; Biological rhythms -- Periodicals; Circadian rhythms -- Periodicals; 571.77
• DOI: 10.3109/07420528.2015.1016613
• Languages: English
• ISSNs: 0742-0528

Abstract

<abstract> <title> <x xml:space="preserve">Abstract</x> </title> <p>Since there is less movement during sleep than during wake, the recording of body movements by actigraphy has been used to indirectly evaluate the sleep–wake cycle. In general, most actigraphic devices are placed on the wrist and their measures are based on acceleration detection. Here, we propose an alternative way of measuring actigraphy at the level of the arm for joint evaluation of activity and body position. This method analyzes the tilt of three axes, scoring activity as the cumulative change of degrees per minute with respect to the previous sampling, and measuring arm tilt for the body position inference. In this study, subjects (<italic>N</italic> = 13) went about their daily routine for 7 days, kept daily sleep logs, wore three ambulatory monitoring devices and collected sequential saliva samples during evenings for the measurement of dim light melatonin onset (DLMO). These devices measured motor activity (arm activity, AA) and body position (P) using the tilt sensing of the arm, with acceleration (wrist acceleration, WA) and skin temperature at wrist level (WT). Cosinor, Fourier and non-parametric rhythmic analyses were performed for the different variables, and the results were compared by the ANOVA test. Linear correlations were also performed between actimetry methods (AA and WA) and WT. The AA and WA suitability for circadian phase prediction and for evaluating the sleep–wake cycle was assessed by comparison with the DLMO and sleep logs, respectively. All correlations between rhythmic parameters obtained from AA and WA were highly significant. Only parameters related to activity levels, such as mesor, RA (relative amplitude), VL5 and VM10 (value for the 5 and 10 consecutive hours of minimum and maximum activity, respectively) showed significant differences between AA and WA records. However, when a correlation analysis was performed on the phase markers acrophase, mid-time for the 10 consecutive hours of highest (M10) and mid-time for the five consecutive hours of lowest activity (L5) with DLMO, all of them showed a significant correlation for AA (<italic>R</italic> = 0.607, <italic>p</italic> = 0.028; <italic>R</italic> = 0.582, <italic>p</italic> = 0.037; <italic>R</italic> = 0.620, <italic>p</italic> = 0.031, respectively), while for WA, only acrophase did (<italic>R</italic> = 0.621, <italic>p</italic> = 0.031). Regarding sleep detection, WA showed higher specificity than AA (0.95 ± 0.01 versus 0.86 ± 0.02), while the agreement rate and sensitivity were higher for AA (0.76 ± 0.02 versus 0.66 ± 0.02 and 0.71 ± 0.03 versus 0.53 ± 0.03, respectively). Cohen's <italic>kappa</italic> coefficient also presented the highest values for AA (0.49 ± 0.04) and AP (0.64 ± 0.04), followed by WT (0.45 ± 0.06) and WA (0.37 ± 0.04). The findings demonstrate that this alternative actigraphy method (AA), based on tilt sensing of the arm, can be used to reliably evaluate the activity and sleep–wake rhythm, since it presents a higher agreement rate and sensitivity for detecting sleep, at the same time allows the detection of body position and improves circadian phase assessment compared to the classical actigraphic method based on wrist acceleration.</p> </abstract>