Reduced multiscale complexity of daily behavioral dynamics in autism spectrum disorder

Abstract

Aim: Autism spectrum disorder (ASD) is difficult to diagnose objectively due to its heterogeneous and complex manifestations. This study aimed to objectively characterize the behavioral phenotypes of ASD children by exploring the multiscale behavioral dynamics.

Methods: We applied behavioral organization (BO) and multiscale sample entropy (MSE) analyses to physical activity data collected from ASD and typically developing children, using wearable monitors in their daily life. We also examined their correlation with auditory startle response measures and clinical questionnaires, including the Social Responsiveness Scale (SRS) and the Strengths and Difficulties Questionnaire (SDQ).

Results: A significant decrease in MSE at timescales longer than 6 min was observed in ASD children, suggesting decreased irregularity or unpredictability, potentially linked to repetitive behaviors or stereotyped patterns commonly observed in ASD. Additionally, an increase in MSE positively correlated with prepulse inhibition levels, indicating its relationship with sensorimotor gating. Moreover, the observed significant negative correlation with the total difficulty score of SDQ substantiates MSE's potential as an objective metric for assessing general mental health problems associated with ASD.

Conclusion: Multiscale analysis enhances the understanding of ASD's behavioral dynamics, providing valuable metrics for real-world assessments.

Keywords: autism spectrum disorder; complexity; multiscale dynamics; physical activity.

Figure 1

Flowchart of study population. ASD, autism spectrum disorder; TD, typically developing.

Figure 2

Behavioral organization analysis of daily physical activity data for typically developing (TD) children and those with autism spectrum disorder (ASD). (a) Cumulative distributions of resting‐period durations for TD (blue circles) and ASD (red rectangles) groups. Error bars show standard error of the mean. Straight lines are used as visual aids, representing the overall mean values; $\gamma$ = 1.18 for ASD: $\gamma$ = 1.11 TD. (b) Similar to (a), but for active period durations.

Figure 3

Multiscale sample entropy analysis of daily physical activity data. (a) Illustration of coarse‐graining processes. (b) Graph showing sample entropy values (sampEn) plotted against scale factor for typically developing (TD) (blue circles) and autism spectrum disorder (ASD) (red rectangles) groups. Error bars represent standard error of the mean. Asterisk (*) indicates a statistically significant group difference at the 0.05 level.

Figure 4

Significant correlations between multiscale sample entropy (scale factor = 18) and ASR measures and baseline questionnaires. Linear regression lines are depicted for each panel. ASD, autism spectrum disorder; SDQ, Strengths and Difficulties Questionnaire; TD, typically developing.