Note: Most Internet Explorer 8 users encounter issues playing the presentation videos. Please update your browser or use a different one if available.

Rating Scale Measures Are Correlated with Automated Video Tracking of Behaviors of Children On the Autism Spectrum

Thursday, 2 May 2013: 09:00-13:00
Banquet Hall (Kursaal Centre)
I. L. Cohen1, T. R. Gomez1, S. Y. Kim1, B. Z. Karmel2 and J. M. Gardner1, (1)New York State Institute for Basic Research in Developmental Disabilities, Staten Island, NY, (2)Infant Development, NYS Institute for Basic Research in Developmental Disabilities, Staten Island, NY

Children with Autism Spectrum Disorder (ASD) display problems with socialization and show stereotyped movement patterns.  We have clinically observed that such children often track the periphery of an unfamiliar room, and  have seen similar behaviors in at-risk toddlers. While typically assessed with rating scales, social and repetitive behaviors also can be quantified with physical parameters (e.g., time spent in a target area, distance from target, speed of movement, etc) that can be automatically computed.  Automated tracking has advantages over subjective ratings in terms of reliability and amount of information provided. It potentially is important for assisting with diagnosis, informing animal models of ASD, and providing objective measures of treatment intervention.  However, its validity for ASD has not been examined.


To examine the degree to which automated tracking data correlate with rating scale measures of ASD.


Children were observed during free play for 3 minutes before and 3 minutes after testing with the Autism Diagnostic Observation Scale – Generic (ADOS-G) in a large room (3.18 m by 4.85 m) with toys placed on the floor and table.  The parent was seated in the northwest corner and asked to complete the Aberrant Behavior Checklist (ABC).  NOLDUS Ethovision XT software provided X-Y coordinates of the child derived from a centrally-located ceiling-mounted camera.

To date, 19 observations from 18 children 3 to 14 years (mean (SD) = 7.2 (3.2) years) have been obtained.  Diagnoses were: 14 ASD, 2 anxiety disorder, 1 ADHD; and 1 developmental delay.  Ratings (completed by parent) included: PDD Behavior Inventory (PDDBI); ABC (Sansone et al (2011) factors); parent interview with Vineland Adaptive Behavior Scales (VABS); and clinician ratings using Gotham et al (2007) scoring system for the ADOS-G.  Ethovision measures included: mean distance (meters) from parent, from center of room, and from periphery (walls 90⁰ and 180⁰ away from parent); percent time near parent or periphery; speed (km/hour) of movement and rate of circling the room (responses/min).  Correlation matrices were computed and p <0.05 was adopted for this exploratory study.


The greater the distance from parent,  the worse the scores for: PDDBI Sensory, Social Approach, Expressive Language, and Learning, Memory and Receptive Language; ABC Stereotypy and Social Withdrawal; ADOS-G  Social Affect; and VABS Communication, Daily Living Skills, and Socialization domains (Pearson rs: 0.47-0.67).

The greater the time spent in periphery of room, the worse the scores for: PDDBI  Sensory, Aggression, Social Approach, and Expressive Language Competence; ABC  Social Unresponsiveness, Stereotypy, Hyperactivity, and Social Withdrawal; and VABS Communication, Daily Living Skills, Socialization and Motor Skills domains (Pearson rs: 0.48-0.67).

The greater the rate of circling, the worse the scores for: PDDBI  Sensory, Ritualisms, Social Pragmatic Problems, Arousal, Specific Fears, Aggression, and Social Approach; ABC Irritability, Social Unresponsiveness, Stereotypy, Hyperactivity, and Social Withdrawal; and VABS  Motor Skills domains (Pearson rs: 0.46-0.77).


These results are encouraging and suggest that automated tracking of social distance and repetitive behaviors yields valid information for assessing children with autism.  We are collecting more data (including unaffected controls) to confirm these observations.

| More