Aggressive, self-injurious, and disruptive behaviors are a common referral concern among individuals with Autism Spectrum Disorder (ASD). These behaviors limit access to educational services and community resources, prevent the development of adaptive behavior, restrict opportunities for socialization, and drastically affect quality of life for families and long-term outcomes for the individuals. There is a critical need fr direct, continuous, and objective measures of these behaviors that are sensitive to treatment-related change. The central innovation of the proposed project is the first, automated approach to directly measure the frequency and intensity of self-injurious, aggressive, and disruptive behaviors in individuals with ASD. This project will establish the accuracy of the automated measurement, evaluate whether it is sensitive to change and improves prediction of pre- post-treatment response over and above commonly-used parent report measures, and explore the feasibility of parents using this approach to gather data on challenging behaviors in the home. These forms of problem behavior involve intensive, characteristic movements. Our Automated Measurement (AM) approach captures these movements and their amplitude (intensity) via accelerometers attached to the wrists and ankles. We apply activity recognition algorithms to the sensor data to automatically segment problem behavior events and classify them as aggression, self-injury or disruption. We will evaluate this measurement approach in a clinic sample of individuals with ASD via three aims. (1) We will determine the agreement between AM of aggression, self-injury, and disruption and live scoring by trained human observers in 24 individuals with ASD referred for treatment for problem behavior. (2) We will evaluate whether AM detects change in the rate and intensity of these behaviors from pre- to post-treatment in 12 individuals with ASD (as determined by live scoring), and whether it is more sensitive to change than parent-report measures commonly used in research and clinical trials. (3) We will demonstrate the acceptability and fidelity of data collection via accelerometers among 10 caregivers and their children with ASD in a home setting, and we will gather preliminary evidence on the agreement between automated measurement of problem behavior from home accelerometry recordings and parent-collected observational data gathered during the in-home deployment. Our use of accelerometers and computational analysis represents an innovative approach to address the measurement challenges raised in the literature. Unlike caregiver questionnaires or clinician ratings, accelerometers directly and continuously measure body movements that are generated by the behaviors of interest. When combined with computational analysis, they produce measurements that are objective. This portable and low-cost measurement approach may open up potential to scale assessment and treatment across clinical and home settings, to assess generalization of treatment gains to everyday life, and enable new research on the etiology, neurobiology, and environmental influences of challenging behaviors in ASD.
The proposed project will develop a novel, automated approach to directly measure the frequency and severity of aggressive, self-injurious, and disruptive behaviors in individuals with Autism Spectrum Disorder (ASD), addressing a critical need for direct and objective measures of these behaviors that can be used to assess the effectiveness of treatments. The project will also explore the feasibility of parents using this approach to gather data on challenging behaviors in the home. Our portable, low-cost measurement approach will open up potential to measure problem behaviors in natural settings such as homes, to assess how treatment gains generalize to everyday life, and ultimately to enable new research on challenging behaviors in ASD, a critical public health concern.
