Day-to-Day Variability in Behavior Via Smartphone Data Collection from Caregivers of Children with Autism
Objectives: The goal of the present study was to validate single questions on a smartphone completed by a caregiver of a child with autism across 8-weeks, compared to gold-standard, paper questionnaires. The second goal was to determine a method to accurately capture day-to-day fluctuations in caregiver reports about their child’s behavior on the smartphone.
Methods: 20 caregivers completed single questions about their child’s mood, irritability, disruptive behavior and anxiety on an 8-point scale across 8-weeks with a modified version of the JAKE™ application. Children were diagnosed with autism and were 5-13 years of age. Consistency between caregiver report on a smartphone versus standardized paper questionnaires was analyzed with mixed ordinal logistic regression models. A total variance measure for each smartphone question was calculated by combining the probability of responses on each day, so that total variance values close to 1 reflected high variability of responses to each number on the 0-7 scale across 8-weeks, whereas values close to 0 reflected little variation in smartphone responses. Lastly, we re-calculated the total variance with N days at the beginning and end of the study to determine how many days of data was necessary to accurately capture variability.
Results: All smartphone questions addressing irritability, negative mood, disruptive behavior, and anxiety were significantly associated with the standard caregiver questionnaires and related subscales on the Aberrant Behavior Checklist (ABC), Child Behavior Checklist (CBCL) and the Positive and Negative Affect Schedule (PANAS) (p’s < 0.05). Maternal education influenced caregiver reporting, such that the more educated the mother, the more severe children were rated on the smartphone, specifically irritability and disruptive behaviors. We found that greater variability on the smartphone questions that targeted mood, irritability and disruptive behavior across the eight weeks corresponded to more severity on the standard paper questionnaires listed above. Interestingly, increased variance on one smartphone question was associated with increased variance on another. Lastly, 7-10 days at the beginning and 7-10 days at the end of the study were able to capture similar amounts of variance as 8-weeks worth of data.
Conclusions: These findings suggest that the smartphone questions were consistent with standardized measures and thus in the future smartphone questions may be used to save time for studies in lieu of longer questionnaires. In addition, these findings suggest that it may be sufficient to deploy cell phones for a week at the beginning and end of an intervention study in order to reliably capture change. Together the data support the use of smartphones as a flexible method to gather data about a child with autism outside of a clinical setting.