Validation of ACT24 Version 2.0 for Estimating Behavioral Domains, Active and Sedentary Time

Area of Science:

• Physical Activity Measurement
• Digital Health Tools
• Behavioral Epidemiology

Background:

• Accurate assessment of physical activity and sedentary behavior is crucial for public health research.
• Traditional methods like direct observation are resource-intensive, while objective measures like accelerometers may not capture activity context.
• Mobile-based tools offer a promising avenue for scalable and context-aware physical activity assessment.

Purpose of the Study:

• To evaluate the criterion validity of the Activities Completed over Time (ACT24) mobile tool.
• To compare ACT24's estimation of sedentary versus active time against the activPAL inclinometer.
• To assess ACT24's accuracy in estimating time spent in various activity domains against direct observation (DO).

Main Methods:

• 53 participants wore an activPAL device for 7 days and completed three automated ACT24 recalls.
• A subset of 24 participants underwent two 3-hour video-recorded direct observation (DO) sessions.
• Paired t-tests, Spearman correlations, kappa statistics, and linear mixed models were used for data analysis.

Main Results:

• ACT24 showed minimal mean differences compared to activPAL for sedentary (1.9%) and active time (2.2%).
• Correlations between ACT24 and activPAL were strong for both sedentary (R=0.61) and active time (R=0.65).
• Substantial agreement (κ=0.63-0.79) was found for leisure, work, and shopping domains; moderate for transportation (κ=0.49); fair for personal care (κ=0.24) and household (κ=0.33).

Conclusions:

• The ACT24 tool demonstrates accuracy for group-level estimation of active and sedentary time.
• Domain-specific agreement with ACT24 is generally higher for frequently reported and longer-duration activities.
• ACT24 shows potential as a valid tool for assessing physical activity patterns in population studies.