Validating functional data analysis measures from 24-h actigraphy data

ABSTRACT

The timing of sleep and wakefulness are under the influence of the circadian pacemaker in the suprachiasmatic nucleus, and the set-point for the circadian pacemaker is revealed by the time of onset of the secretion of melatonin under dim light conditions (DLMO). Measuring DLMO is time intensive and requires biochemical assays; hence it would be desirable to develop simpler alternatives to DLMO measurement that would otherwise provide similar information about health and illness. Continuous, multi-day measurement of actigraphy can be processed to reveal underlying features such as the time of greatest activity (acrophase), or the time of least activity (bathyphase). In this paper, we explore the calculation of the acrophase and bathyphase using a statistical approach called functional data analysis (FDA), and validated the FDA-calculated bathyphase by correlating it with morningness–eveningness self-reports from 47 depressed adults.

KEYWORDS:

• Actigraphy
• functional data analysis
• bathyphase
• acrophase
• validity

Acknowledgments

The authors are grateful to Drs. Meredith Rumble, Ruth Benca for allowing us to use actigraphy data that was in part collected by their team. Dr. Doug Case assisted in the analysis of data other than the actigraphy data.