Improving Autism Outcome Measures: An Integrated Home and Clinic Protocol with Novel Technologies

Friday, May 13, 2016: 11:30 AM-1:30 PM
Hall A (Baltimore Convention Center)
R. M. Jones1, C. Carberry1, A. Hamo1, H. Rao2, U. Gupta2, A. Albin2, R. Pawar2, I. Kleckner3, O. O. Wilder-Smith4, M. S. Goodwin4, M. Clements2 and C. Lord1, (1)Weill Cornell Medical College, White Plains, NY, (2)Georgia Institute of Technology, Atlanta, GA, (3)University of Rochester Medical Center, Rochester, NY, (4)Northeastern University, Boston, MA
Background: Sensitive, standardized, reliable outcomes for clinical trials attempting to change behavior in children and adults with Autism Spectrum Disorder (ASD) have been limited to date. Previous clinical trials often rely on a single behavior to quantify change (Findling et al., 2014) or determine change in one context (i.e. solely parent report or assessments only in the clinic) or use a measure with limited capability for replication (Bolte & Diehl, 2013), thus reducing the scope of the research. Outcomes in clinical trials that combine multiple domains to measure behavior and use multiple contexts to collect data offer the greatest success of capturing change in individuals with ASD.

Objectives: 1) To identify and evaluate social communication and physiological arousal measures that may be sensitive to change; 2) To determine an integrated clinic, smartphone/online and home protocol that is cost effective, unobtrusive and maximizes participant retention. No treatment intervention is part of this study, as the aim is to test the scalability of measurements that track behavior over time.

Methods: 16 children and adolescents with ASD (5-17 years of age), with at least 2-3 word phrases, completed a 1-week protocol. Data collection is ongoing in 20 children and adolescents in an 8-week protocol measuring behavior in both the clinic and the home. In both settings, electrodermal activity (EDA) data were collected via wrist sensors (Q sensors) and natural language data were recorded via LENA digital language processors. Caregivers completed momentary questions about their child’s behavior on a smartphone.

Results: All families completed the protocol, including answering daily smartphone questionnaires and using the LENA devices and Q sensors in their home. As we predicted, there is substantial variation across children in behavior across contexts in amount of language produced and occurrence of problem behaviors. Novel voice detection procedures correctly identified the child’s voice at approximately 90% accuracy as well as affective vocalizations (laughing), providing an automated alternative for time intensive transcriptions and replacing current LENA supported algorithms not suited for children above 5 years of age. As predicted, EDA data from the clinic visit demonstrates increased physiological arousal during mildly frustrating tasks in 63% of the participants. Home EDA data are considerably more variable and changes in EDA do not consistently correspond to increased screaming detected from the LENA recordings, confirming established findings that EDA is complicated to interpret without additional contextual information. Parent momentary reporting demonstrates some consistency with standardized questionnaires (negative mood PANAS) and events identified on the LENA, but are also variable across time and parents.

Conclusions: Caregivers of children and adolescents with autism are motivated to use novel technologies in a daily home and clinic protocol to measure their child’s behavior. Children’s behavior does vary across contexts and there is variability in parental styles in reporting their child’s behavior. The research provides an important foundation to inform future outcome protocols that measure caregiver-reported, behavioral, and physiological changes in children undergoing medical and behavioral interventions.