A Multi-Site Study of Noldus Ethovision XT Video Tracking of Children with and without Autism: Results from the ABC-CT

Background: Parent-child interactions are a naturalistic method of measuring child behaviors and response to treatment. Manual coding is conducted to quantify child behaviors, but this method is time-intensive and subjective. Video tracking offers an automated alternative to coding, yielding quantitative, objective measures of behaviors. Noldus EthoVisionXT (EVXT) was developed to quantify animals’ movements, and was recently applied to children with autism spectrum disorder (ASD) and a non-ASD group during a parent-child free play task (PCFP) (Cohen et al., 2014). As the only study of its kind, further research is needed to assess the validity and reliability of this method.

Objectives: 1) Examine correlations of EVXT outcomes with child characteristics and clinical measures; 2) Determine relations among EVXT outcomes across timepoints; 3) Assess effects of group and timepoint on EVXT outcomes for school-aged children with ASD and TD during a lab-based PCFP.

Methods: As part of the Autism Biomarkers Consortium on Clinical Trials (ABC-CT), a multi-site longitudinal study, school-aged children (baseline mean Age = 8.73 years, range = 6.02-11.53 years; baseline mean FSIQ = 101; range = 60-150) with ASD (N= 161) and TD (N=64) and their caregivers participated in three 6-minute PCFPs (baseline, week 6, week 24). EVXT software generated durations, latencies, and frequencies in core regions of interest (ROIs; Periphery, Caregiver, and Activity), and Total Distance Moved and Mean Distance to Caregiver. Group differences, longitudinal change, and associations of EVXT outcomes with child characteristics (age, gender, IQ) and clinical measures of hyperactivity (CASI-5), autism severity (ADOS-2, SRS), and social-communication skills (Vineland Adaptive Behavior Scales, PDD Behavior Inventory) were explored.

Results: At Baseline, the ASD group spent more time in the Center Activity ROI (F=4.74, p=.03) between the Parent and Table ROIs, whereas the TD group spent more time at the Table Activity ROI (F=3.57, p=.06), across from the Parent ROI. Baseline analyses show associations of EVXT outcomes with parent report outcomes. Lower expressive/receptive social communication ability (PDDBI) correlated with more time in the Periphery ROI for the ASD group (r = -0.18614, p=.023). Higher hyperactivity ratings (CASI-5) correlated with greater Total Distance Moved for the TD group (r= 0.29435; p=.024), and more time in core ROIs for the ASD group (r=0.17464; p= .033). EVXT outcomes across timepoints were positively correlated (Baseline/Week 6: r’s = .216 - .469, p’s = .009 – <.0001; Baseline/Week 24: r’s = .175 – .562, p’s = .035 - <.0001). Analyses of longitudinal change and the effects of site, gender, age and IQ on EVXT outcomes are underway.

Conclusions: This study is the first to use EVXT in a multi-site, longitudinal study of children with ASD and TD. Preliminary analyses of baseline EVXT outcomes indicated different movement patterns for ASD and TD children. Movement patterns detected via EVXT were correlated with individual differences in parent-reported behavior, including social communication ability and hyperactivity. These results suggest that EthoVision is a promising method for automatically tracking individual differences in children’s movements that correlate with clinical features. Methodological considerations for applying EVXT in longitudinal clinical trials are discussed.