Annual Meeting of the NCI Cohort Consortium (Abstract Submission): Submission #4

Submission information
Submission Number: 4
Submission ID: 127219
Submission UUID: 55ea74f5-d525-466a-b132-ce1fc5e73246

Created: Tue, 09/10/2024 - 17:55
Completed: Tue, 09/10/2024 - 17:57
Changed: Tue, 09/10/2024 - 17:57

Remote IP address: 10.208.24.118
Submitted by: Anonymous
Language: English

Is draft: No





Lightning Talks Abstract
------------------------






Presenter's First Name: : Sidney









Presenter's Last Name:: Donzella









Title (eg: professor, assistant professor, chair, etc):: PhD Candidate









Degree(s): MPH









Contact Email:: sdonzell@fredhutch.org









Organization:: University of Washington; American Cancer Society 









Project Title:: Sleep measured by accelerometry: Comparing the performance of objective sleep algorithms









Additional Authors:
1. First Name: Asuka
   Last Name: Ishihara
   Degree(s): PhD
   Organization: NIDDK
2. First Name: Antoinette
   Last Name: Rabel
   Degree(s): PhD
   Organization: NIDDK
3. First Name: Ranganath
   Last Name: Muniyappa
   Degree(s): PhD
   Organization: NIDDK
4. First Name: Kong
   Middle Initial: Y
   Last Name: Chen
   Degree(s): PhD
   Organization: NIDDK
5. First Name: Samuel
   Middle Initial: R
   Last Name: LaMunion
   Degree(s): PhD
   Organization: NIDDK










Abstract::
Background Various algorithms and software are available to score sleep data collected from accelerometer-based wearable sensors. However, little is known about how the combination of algorithms and software may influence the sleep estimates. The aim of this study was to investigate the agreement of sleep constructs scored with different sleep bounding algorithms on two commonly used software platforms (i.e., ActiLife and GGIR). Methods Thirty-five healthy volunteers were invited to wear an ActiGraph wGT3X-BT device for 7 consecutive days and complete a sleep diary. Objective sleep was scored using the count-based Cole-Kripke sleep scoring algorithm and the following sleep bounding algorithms: GGIR heuristic of the distribution in the change in Z-axis angle (HDCZA) with open-source counts, GGIR HDCZA zero crossing counts, GGIR open-source counts with sleep diary, ActiLife Tudor Locke. We included participants who had 16 hours of device wear/day and 4 days of valid device wear. We used intraclass correlation coefficient (ICC) with 95% confidence intervals to assess the agreement of objective sleep constructs with the sleep diary. Results Twenty-nine participants were included. ICC estimates for time asleep and time awake ranged from moderate to good agreement (ICC range 0.53-0.86) for all algorithms compared to the sleep diary. Compared to the sleep diary, all algorithms showed poor agreement for sleep duration (ICC range -0.64-0.47). Conclusion Sleep constructs scored using different algorithms or on different software platforms may not be generalizable to one another.  Researchers should use caution when comparing objective sleep measures across studies that used different sleep scoring procedures.