Inhome: A Multimodal Bio-Behavioral Data Capture System for Autism Research

Background: The majority of existing research in Autism Spectrum Disorder (ASD) relies on parent- or informant-report measures and/or behavioral assessments collected at a limited number of time points, often in unfamiliar settings (laboratory, clinic). Recent advances in commercially available multimodal recording technology make it possible to move beyond reliance on both approaches and collect intensive longitudinal video and physiological data in home-based settings. These technologies have the potential to provide greater statistical power, measurement precision and sensitivity, and enhanced ecological validity.

Objectives: (1) Develop a multimodal bio-behavioral data capture system to unobtrusively, efficiently, and accurately record and analyze behavior and physiology in individuals with ASD in home settings for up to 1 contiguous month. (2) Evaluate usability, feasibility, and data quality of our system in a sample of children with ASD and their families.

Methods: We developed inHome (In-Home Observation Measurement Equipment), an integrated system consisting of commercially available cameras, laptop computers, electrodermal activity sensors (Q Sensor), sleep sensors (AMI), and tablet computers that provide video review and annotation capabilities. inHome is currently configured to collect up to 30 days (5 hours per day) of continuous video, audio, physiological, physical activity, sleep, and informant-report data that is automatically synchronized and recorded to a central repository.

Results: To-date, 10 children with ASD (age 4-14 years) and their families have completed trialing the system in their homes for 2-4 weeks, for a total of 27 weeks of data collection, and approximately 900 hours of video, audio, physiological, and physical activity data. Installation of the system in a family’s home and training on its use can be completed in less than 90 minutes. Parent feedback from post-deployment interviews indicates that the system is easy-to-use, unobtrusive, and able to capture data that they believe is representative of their children’s behavior.

Conclusions: We have successfully developed a novel system for collecting intensive longitudinal multimodal data on behavior, physiology, physical activity, and sleep, in individuals with ASD in home settings. The data obtained thus far represents a first-of-its-kind, long-term intensive longitudinal dataset of bio-behavioral data from children with ASD in home settings. In addition to complementing traditional survey and lab-based measures, enabling parents to efficiently gather high quality quantitative assessments of their children’s behavior and physiology in the home over time could enhance intervention and clinical trials by establishing more sensitive and detailed outcome measures. It may also facilitate more basic research to the extent that our systems’ measures can be associated with other biological indices (genetic, metabolic, proteomic, immunologic, neurologic, psychiatric, etc.) obtained from a large number of individuals on the autism spectrum. Our demonstration will include opportunities to try out and interact with the inHome system, as well as review examples of data collected from our end-user deployments.