|Real-Time Sensing Devices: An Opportunity to Deploy Behavioral Theory in Health Interventions|
|Saturday, May 26, 2018|
|3:00 PM–3:50 PM |
|Manchester Grand Hyatt, America's Cup A-D|
|Area: CBM/EAB; Domain: Translational|
|Chair: Vincent Berardi (Chapman University)|
|Discussant: Melbourne F. Hovell (SDSU School of Public Health)|
Real-time sensing devices such as fitness trackers and the Internet of Things are beginning to enable the continuous assessment of a large range of health behaviors as well as the physiological/environmental contexts in which they occur. This stands in contrast to the historical, non-dynamic, episodic assessments (e.g surveys, EMAs) typically implemented in health-behavior interventions. Real-time sensing technologies increase the precision of both behavioral observations and the deployment of operant contingencies, thereby more faithfully approximating the experimental conditions in lab-based studies. Insights from applied behavior analysis can thus inform health intervention designs, which is expected to increase their efficacy. This symposium presents results from two large health behavioral trials that used real-time sensing technology and operant contingencies to reduce tobacco and increase physical activity behaviors. The first study will focus on the use of real-time air particle monitors to increase the latency between tobacco use responses and the presentation of contingent, aversive stimuli. Longitudinal methodologies for detecting avoidance versus escape behavior in response to aversive stimuli will also be discussed. The second study will describe the use of wrist-worn accelerometers to explore differential effects of varying schedules of financial reinforcement (e.g. fixed vs. variable magnitude/ratio) to promote daily physical activity over one year among healthy, inactive adults. The discussion will focus on the implications of these technological advances and the opportunities for automated shaping routines designed to engender healthy, targeted behavior.
|Instruction Level: Intermediate|
|Keyword(s): auto-shaping, health behavior, real-time sensing|
Effects of Continuous ReinforcementVersus Variable Magnitude and Probability Reinforcement on Adults' Physical Activity
|MARC A. ADAMS (Arizona State University)|
A secondary analysis of data from one arm of an RCT will be presented in order to explore the effects of six different reinforcement stages, each characterized by a different positive reinforcement schedule, on participants' MVPA. Participants (N=187) wore an ActiGraph GT9X daily (32,632 observations) in a one-year intervention. They were prescribed MVPA goals (i.e., MVPA bout minutes) and could earn rewards daily for meeting goals. As participants met goals, they progressed through four stages, transitioning from one to the next after meeting 24 goals: 1) continuous fixed reinforcement ($1 per goal met), 2) continuous variable magnitude ($0.25 to $2.50/goal met, all goals earned a reward), 3) variable magnitude and probability ($0.25 to $2.50/goal, 8% of goals earned $0), 4) variable magnitude and probability ($0.25 to $2.50, 17% of goals earned $0). Rewards denominations were randomized and total amounts within each stage summed to $24. Participants were paid immediately via electronic gift cards after earning $5.00. We averaged cumulative MVPA bout minutes/day (bm/d) over all participants on each day in each stage to investigate potential differences between stages. Rate of growth in cumulative MVPA bm/d increased across stages, with Stage 1 (continuous reinforcement) having the smallest slope. Linear regression indicated that a slope of 23.8 bm/d in Stage 1 increased to 30.1 bm/d in Stage 2 (? = 6.3, p<.001). The slope increased by an additional 2.1 bm/d at Stage 3 (p=0.23) and increased by 12.5 bm/d at Stage 4 (p<.001). When repeating this analysis with only the highest-performing participants, the ? between Stage 1 and Stage 2 was smaller (2.8 bm/d, p<.001) and slopes for other adjacent pairs of stages were more similar. Progression through both variable magnitude and variable probability reinforcement schedules resulted in higher MVPA than observed with continuous fixed reinforcement alone, with leaner probability schedules producing the largest amount of MVPA accrual. Variance was not constant across subjects and was smallest for the highest performing participants. Implications for behavioral maintenance will be discussed.
Project Fresh Air: Real-Time Feedback to Encourage Smoke-Free Home
|JOHN BELLETTIERE (University of California San Diego ), Suzanne Hughes (San Diego State University, Center for Behavioral Epidemiology and Community Health), Neil Klepeis (San Diego State University, Center for Behavioral Epidemiology and Community Health), Sandy Liles (San Diego State University, Center for Behavioral Epidemiology and Community Health), Benjamin Nguyen (San Diego State University, Center for Behavioral Epidemiology and Community Health), Marie Boman-Davis (National University)|
Exposure to fine particulate matter in the home from sources such as smoking, cooking, and cleaning may put residents, especially children, at risk for detrimental health effects. A trial was conducted from 2011 to 2016 to determine whether real-time feedback in the home could reduce fine particle levels in homes with smokers and children. Monitors were installed in each subject's home and air particle measurements were collected on a nearly-continuous basis over the course of several months. A subset of homes, designated as the treatment group, received aversive visual and auditory feedback (yellow/red lights and tones) when air particle concentrations exceeded a threshold, representing a punishing contingency. A major feature of this study is a very small latency, defined as the temporal interval between a behavior and the presentation of a contingent consequence. This characteristic is known to strengthen behavioral contingencies. In a separate arm, punishing stimuli were augmented with reinforcing contingencies, with participants being provided with gift cards with a value contingent upon the duration of contiguous low-level measurements. The required duration to activate reinforcement was tailored to participants' baseline measurements. This real-time behavior intervention was successful in reducing both the number of particle generating events and the mean daily particle concentration. Moreover, it demonstrates the ability of real-time technology to more faithfully adhere to applied behavioral principles than a traditional approach and to allow for a level of personalization that is often lacking in behavioral interventions. As real-time sensing technology becomes even more ubiquitous, this capacity will grow and allow other features, such as behavior shaping, to be incorporated into interventions. We expect that this development will have major implications for improving the efficacy of health behavior interventions, particularly concerning the maintenance of healthy behaviors.