Physiological Monitoring during PEERS®: A "Culture-Free" Method of Understanding Intervention Response

Background: Adolescents with ASD are at an increased risk of anxiety disorders (White et al., 2009) versus typically developing individuals.  Arousal from the autonomic nervous system is the somatic component of anxiety (Vitasari et al., 2011) and is experienced by an increase in heart rate, increase in electrodermal activity (Critchley, 2009), and shortness of breath.  In addition to physiological arousal, emotions are expressed behaviorally through facial expressions, body gestures, and speech. No study has examined electrodermal activity (EDA) as a base to understand anxiety and arousal over the course of intervention in ASD, even though objective, universal measures would be beneficial to understanding how and why certain interventions are conducive for positive outcomes.

Objectives: To examine the relationship between presence of physiological arousal, expression of this emotion through facial features, and self-awareness of anxiety among adolescents with ASD who participated in the PEERS® intervention.  The current study aims to address whether physiological arousal can serve as a “culture-free” measure of intervention response.

Methods: For the present study, 13 adolescents (11 to 18 years-old) with ASD were drawn from a larger ongoing study over the span of 1.5 years. All participants had an IQ greater than 70 on KBIT and diagnoses were confirmed with the ADOS. Adolescents participated in the Program for the Education and Enrichment of Relational Skills (PEERS: Laugeson & Frankel, 2010). Measures that were taken during each of the 14 intervention sessions included (1) 30 minutes of electrodermal activity (Q-Sensor wristband: Affectiva, Inc.), (2) self-report of anxiety (STAI: State Trait Anxiety Inventory: Short Form: Marteau & Bekker, 1992), and (3) video coding of facial affect (The EigenFace method: Turk & Pentland, 1991).

Results: Regression analysis indicated a significant relationship between maximum EDA during each PEERS session and the average score for self-reported anxiety, F(9,89) = 2.16, p< .05. The EigenFace method (for identification of facial affect from facial images), followed by a machine learning algorithm (for image classification), was only able to predict the EDA arousal states at a slightly better than chance accuracy, 57%. The image clustering technique (k-means) did not provide a clear distinction between facial affect images that corresponded to highest and lowest EDA in the PEERS sessions for each participant.

Conclusions: EDA and self-report data shows the presence of physiological arousal and anxiety in adolescents with ASD during PEERS intervention sessions. However, adolescents’ facial expressions were not predictive of their anxiety or arousal. A significant relationship between EDA and self-report of anxiety suggests that adolescents are self-aware of their arousal states, and that EDA shows promise as a “culture-free” method of monitoring arousal during intervention. Future objectives include examining trajectories of EDA response and relations to intervention outcome.