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

Characterization of ASD Onset Using Automated Analyses of Child Vocal Patterns

Thursday, 2 May 2013: 14:00-18:00
Banquet Hall (Kursaal Centre)
15:00

ABSTRACT WITHDRAWN

Background: Detection of autism spectrum disorder (ASD) before behavioral symptoms and delayed developmental trajectories establish is an identified research priority (e.g., IACC, 2010; Lord & Bishop, 2010). Past studies of onset have suggested that the trajectory of ASD onset is not uniform across affected children. There appears to be consensus that about a third of children will experience developmental regression sometime in the first 3 years of life (e.g., Baird et al., 2008; Goldberg et al., 2003; Luyster et al., 2005; Werner et al., 2005); these rates are based on retrospective parent report. More recent studies suggest that parent report may underestimate developmental regression compared to prospective clinician observation (Ozonoff et al., 2010; Ozonoff et al., 2011). It is possible, too, that different intensities of regression exist, such that some are obvious and easy for parents to recognize while others are more subtle and only detected by trained clinicians. Similarly, Pickles and colleagues (2009) suggested that the language delays common to ASD could be masking language losses that would otherwise become apparent if the children developed words on time. A critical gap in understanding ASD onset is the availability of sensitive, reliable tools for detecting patterns of early social communication that predict diagnosis. Automated tools may promote this understanding through providing quantitative, objective information on child behaviors. Automated tools might also help detect subtle changes that existing methods may miss.

Objectives: Identify patterns of social communication development that predict ASD using automated audio analysis of child vocal patterns.

Methods: Vocal patterns of 13 baby siblings of children with ASD (ASD-risk) and 7 infants at-risk for general developmental disabilities (DD-risk) were collected and analyzed prospectively using the Language ENvironment Analysis (LENA) system. LENA uses a small device worn on the child to record child vocalizations and adult words spoken near the child. Day-long vocal samples were collected once a month from age 6 to 18 months. Children were assessed every 6 months with the AOSI, Mullen, and ADOS-T (when walking). ASD risk status was determined at 18 months. Trajectories of frequencies of child vocalizations and child-adult vocal interactions were compared separately for (a) ASD-risk versus DD-risk and (b) children showing ASD concerns at 18 months versus children not showing ASD concerns.

Results: Data collection will be complete in January 2013. Preliminary results indicate that, compared to DD-risk, the ASD-risk group shows significantly lower frequency of child-adult vocal interactions between 6 and 12 months. Children showing ASD concerns show a decreasing trajectory in frequency of vocalizations. Groups did not differ on measures of expressive language at 6 or 12 months.

Conclusions: Preliminary findings suggest children at-risk for ASD show differences in social communication patterns in the first year of life that were not detected with standardized communication or behavioral measures. Future work will follow participants to 36 months for final diagnosis. Larger samples are needed to validate these findings. With better understanding of when and how development takes a course toward ASD, we can improve timing of diagnosis and initiation of interventions.

See more of: Core Deficits I
See more of: Core Deficits
See more of: Symptoms, Diagnosis & Phenotype
| More