A new approach for the analysis of accelerometer data measured on preschool children

Area of Science:

• Pediatric research
• Physical activity measurement
• Biostatistics

Background:

• Accelerometers are common tools for quantifying physical activity (PA) in preschoolers.
• Current data treatment methods lack a standardized 'best practice', hindering consistent analysis.
• Developing a robust method for accelerometer data reduction is crucial for accurate PA assessment.

Purpose of the Study:

• To develop and apply a robust method for reducing preschoolers' accelerometer data.
• To utilize contemporary statistical methods for improved PA data analysis.
• To establish a reliable approach for quantifying physical activity in young children.

Main Methods:

• Recruited 2- to 5-year-old children in Auckland, New Zealand, for 7-day accelerometer monitoring.
• Derived average daily PA rates per second using negative binomial generalized estimating equation (GEE) models.
• Compared derived participant rates with traditional data inclusion approaches.

Main Results:

• Data collected from 78 children over a median of 7 days.
• Daily PA rates varied significantly, with a median of 5.70 counts per second.
• The new method demonstrated improved data retention and provided a continuous PA measure, facilitating multivariable regression analyses.

Conclusions:

• Successfully calculated PA rates for preschoolers' activity description.
• Identified advantages of the new statistical approach, including enhanced data retention and continuous measure computation.
• The proposed method shows promise for future research and warrants further application and refinement.