Characterization of Chronotypes Using the Symbolic Aggregate apprXimation (SAX) on Actigraphy Data

Area of Science:

• Chronobiology
• Data Science
• Public Health

Background:

• Actigraphy monitors human rest/activity cycles, offering insights into sleep-wake behaviors and circadian rhythms.
• High-dimensional actigraphy data presents challenges in storage, processing, and analysis.
• Characterizing chronotypes is crucial for understanding health and behavior.

Purpose of the Study:

• To develop an efficient approach for chronotype characterization using Symbolic Aggregate Approximation (SAX) on large-scale actigraphy data.
• To reduce dimensionality of actigraphy data while preserving essential patterns for analysis.
• To identify distinct chronotype clusters and investigate age-related variations.

Main Methods:

• Applied the SAX algorithm to transform continuous time-series actigraphy data into a symbolic representation.
• Utilized unsupervised clustering on SAX-transformed data from the NHANES database (>10,000 individuals).
• Analyzed chronotype clusters for differences in activity onset, resolution, intensity, and age distribution.

Main Results:

• Identified five distinct chronotype clusters using SAX and clustering techniques.
• Revealed significant age-related chronotype variations, with younger individuals exhibiting delayed sleep onset and wake times.
• Observed distinct activity transition dynamics (winding up/down periods) across different chronotype clusters.

Conclusions:

• SAX is an efficient and effective method for processing large-scale actigraphy data for robust chronotype characterization.
• Chronotype variations are linked to age, with younger populations showing delayed sleep patterns.
• Findings can inform personalized healthcare and public health initiatives related to circadian biology.