Poster #75 - Bluetooth-signal contingent ecological momentary assessments of parent-child interactions: Proof of concept study

• In Event: Poster Session 09

Abstract

Background. Ecological momentary assessment (EMA) is a popular approach to measuring thoughts, emotions and behaviors in parents and youth. EMAs are often deployed on a researcher-defined schedule (i.e., signal contingent) or initiated by participants (i.e., event-contingent). More recent studies have attempted to reduce recall bias and increase the specificity of when EMAs are triggered by relying on real-time passive data such as activity levels and bed and wake times (e.g., just in time adaptive assessments). The current study applies this framework to detect dyadic events (i.e., parent-youth interactions) and trigger adaptive EMA to both parent and youth simultaneously. In this presentation, we propose to present results from a proof-of-concept study that shows the feasibility of this novel approach to assessing dyadic interactions. Method. Data comes from a larger naturalistic observational study of parent-adolescent supportive communication and depression symptoms, in everyday life. See Figure 1 for study procedures. We used the software, Wear-It, to conduct EMAs. In Phase 1, we developed R code to replicate the times when the app triggered proximity surveys, using data from eight of the 38 families in Cohort 1. In Phase 2, we will validate this code in data from eight families currently enrolled in Cohort 2. The eight Phase 1 families were white and resided in northeast US; two of the eight teens were males, and all others female. Age ranged from 12 to 16 years. All parents were mothers. Parents and youth participated in a two week-long naturalistic observational phase during which they always carried study smartphones. We use Bluetooth signals emitted and detected by the smartphones to assess physical proximity and automatically deploy proximity-contingent EMAs about parent-youth interactions when real-time proximity data indicate a high likelihood of interaction. Results. EMA alerts were triggered based on two timers. When a phone registered that a partner phone is nearby based on Bluetooth signals, it started a 15-minute timer to determine if there has been a long enough interaction to trigger a survey. After the completion of the 15-minute timer, a second 2-minute timer was initiated to afford participants time to go about their own ways. A proximity-contingent survey was triggered after the 2-minute timer was complete. Parallel timers for “grace periods” additionally offered some protection against brief disruptions in readings due to technological errors resulting in lapses in Bluetooth signal detection. Figures 2 exemplifies results of Phase I; we were able to successfully replicate the times when proximity-contingent surveys were triggered in Cohort 1 families, indicating that the app is working as designed and providing proof of concept for a proximity-triggered EMA alert. Next, we will compare the estimated trigger times to actual trigger times in Cohort 2 families. Discussion. This highly innovative approach draws on the combined strengths of signal- and event-contingent surveys, and significantly reduces the recall bias in traditional EMAs. The current study uses Bluetooth signals to detect dyadic events and signal multiple family members. This technological framework is a prerequisite for delivering adaptive family interventions in the future.

Authors

• Sunhye Bai, Pennsylvania State University, University Park
  Presenting Author

• Liu Bai, Pennsylvania State University, University Park

• Kelly W. Ngigi, Pennsylvania State University, University Park

• Jonathan Weaver

• James Mundie, The Pennsylvania State University

• Timothy Raymond Brick, Penn State Univeristy