Quantifying the Active Ingredients of Music Therapy: An Analysis of Movement Dynamics between Therapist and Children with Autism

Background: In a randomized controlled trial, we recently showed that 8-12 weeks of music intervention can improve auditory-motor brain connectivity and parent-reported social communication in school-age children with ASD (Sharda et al., 2018). This suggest that collaborative music-making can enhance communication. It has been argued that music is unique in engaging both sensory and motor systems while synchronizing with another person, and that this bears positive effects across domains (Srinivasan & Bhat, 2013, Janzen et al., 2018). Yet we know little about the mechanisms that lead to these effects. Prior work has shown that interpersonal synchrony (coded subjectively) increases in children with ASD over the course of a rhythm intervention (Srinivasan et al., 2015), possibly through enhanced movement coordination. Here we applied a novel, optical flow analysis method to video of therapy sessions from Sharda et al. (2018) to examine dynamic movement amplitude and coordination between child and therapist.

Objectives: 1) To apply a novel, optical flow-based methodology to quantify movement amplitude and coordination between therapist and child; 2) Examine how movement amplitude and coordination may evolve over time; 3) Compare therapist-child movement coordination patterns across music-based intervention and a control play-based intervention.

Methods: Data is available from 51 children with ASD aged 6-12; 26 participated in music-based intervention (MI) and 25 participated in a control play-based intervention (PI), all one-on-one with the same therapist. A preliminary analysis of 10 MI participants is described, with the rest of the sample underway. The duration of therapy was one weekly session (comprising 4 activities) over 8-12 weeks. Specific activities were rotated and reoccurred over the course of intervention. For each participant we selected 1 minute video clips of the same activity (e.g., handheld percussion) occurring in the first third (T1) or last third (T2) of therapy sessions. Movement patterns between child and therapist were analyzed using Correlation Map Analysis (CMA) software. Non-overlapping whole-body plus instrument regions of interest were identified for the child versus therapist. CMA uses a standardized optical flow to assess movement within a given region based on changes in video pixels across video frames. A moving-window filter is applied to calculate an instantaneous correlation between regions across the length of the video (Barbosa et al., 2012).

Results: Average amplitude of motion was calculated for both the child and therapist at each time point. Some dyads exhibited more similar or matched motion amplitude over time (Figure 1). Figure 2 shows that this effect was observed across activities; the mean difference in child-therapist motion amplitude significantly decreased from 3.47 at T1 to .72 at T2 (p = .018).

Conclusions: This novel method provides numerous insights into the movement dynamics of a dyad during intervention. Preliminary results demonstrate increased movement matching over the course of music intervention. We are examining individual differences in this effect in our full sample relative to communication and brain connectivity outcomes. Comparisons of movement dynamics between music-based and play-based therapy may elucidate an active ingredient of music intervention: increased interpersonal motion matching.