Relations Between Computerized LENA Recordings of Conversational Turns and Lab-Based Measures of Social Engagement in Children with Autism Spectrum Disorder

Background:  Social communication impairments are core deficits and prognostic indicators of Autism Spectrum Disorder (ASD).  A key challenge in the design and evaluation of ASD treatments is identifying measures of social communication that are ecologically valid, quantitative, and scalable for large clinical trials. Computerized home-based recordings of social communication, such as conversational turns, are promising alternatives to lab-based measures that require labor-intensive human coding.  However, more research is needed to determine the extent to which such computerized recordings correlate with well-established, lab-based measures of social communication.

Objectives:  Aim one of this study was to examine the correlation between a computerized home-based measure of social communication and a measure of social communication derived from human coding of a laboratory-based parent-child play interaction. Aim two assessed the relationships between expressive language in children with ASD and both home-based computerized measures and lab-based coded measures of social communication. 

Methods:  Data were collected on children with ASD (N=25, Mean Age=4.47 years, Range=2.27-5.98 years; Mean Nonverbal IQ=64.3 +/- 24.6). ASD diagnosis was based on the ADOS-2 and ADI-R. Measures consisted of (1) a computerized measure of social communication collected with the LENA (Language Environmental Analysis) System at home, which automatically recorded adult and child word/vocalizations and conversational turns (Mean: 15.95 recorded hours; Range: 14.8 – 16 hours), (2) reliable human coding of the duration of child social engagement states during a lab-based, parent-child interaction, and (3) a standardized measure of expressive language (Expressive One Word Vocabulary Test).  Parent-child play (8 minutes) was coded for six child-focused engagement states: Unengaged, Onlooking, Object, Person, Supported, Coordinated (Adamson et al. 1998). Using LENA, we calculated the average number of conversational turns during each conversation, the average duration of conversational blocks, and adult word count and duration during conversations. Data collection and coding are ongoing; final results will include additional data.

Results:  Analyses of coded parent-child interactions (N=11) revealed that average parent-child conversational turns during conversational blocks at home (LENA) was significantly negatively related to the proportion of time in unengaged and supported engagement states (r=-0.701, p<0.05; r=-0.69, p<0.05, respectively), and positively related to the proportion of time in coordinated engagement (r=0.75, p<0.01) during lab-based play. Results were the same for average duration of conversational blocks. Adult word count during conversations (LENA) was significantly negatively correlated with proportion of time unengaged (r=-0.648, p<0.05) and positively related to coordinated engagement (r=0.661, p<0.05) during parent-child play. Child expressive vocabulary scores were significantly negatively related to supported engagement (r=-0.658, p<0.05) and positively related to coordinated engagement (r=0.916, p<0.001). Coding is ongoing and additional analyses will compare engagement states, conversational turns, and adult words of minimally verbal and verbal children. 

Conclusions:  This study is one of the first to investigate associations between home-based computerized measures and lab-based coded measures of social communication between a parent and child with ASD. Results suggest that home-based computerized measures of social communication may offer a scalable, reliable, and quantitative outcome measure for use in clinical trials of treatments for social communication impairments in young children with ASD.