Improving Theoretical Considerations via Analysis of Secondary Data from Clinical Trials

A significant portion of my research activities have been associated with two R-01 clinical trials that used mobile technology to encourage healthy behavior in real-world environments.

photo by Cristian Guerrero

TOBACCO CESSATION

Project Fresh Air (PFA; PI: Hovell, SDSU), placed air particle sensors/alarms in the households of smokers (N=298) and used real-time feedback to reduce secondhand smoke exposure.

The theoretical precision of my work can also be seen in an article (Berardi et al., 2018)  from the PFA secondhand smoke trial that I published in the Journal of Biomedical Informatics (Q1) This work described a Markov methodology that allowed participants’ responses to aversive feedback to be classified as either avoidance or escape behavior. Avoidance/escape are well-known operant theory constructs and these findings represents a newfound level of nuance that will inform future intervention design.


PHYSICAL ACTIVITY PROMOTION

WalkIT Arizona (PI: Adams, ASU), had participants (N=512) wear an accelerometer (similar to a FitBit) for one year and provided reinforcement for meeting daily exercise goals. While I have aided with the publication of the main outcomes for these trials (Hovell et al., 2020, Adams et al, 2019, Hughes et al., 2018), my more consequential contributions have been associated with the development of analytic methodologies that use the big data from these studies to describe behavioral dynamics more precisely and to relate these observations to established theoretical considerations.

For example, I recently published novel analyses from the WalkIT trial that detailed differential intervention effects as a function of the frequency and magnitude of the reinforcement that participants received for meeting daily exercise goals (Berardi et al. 2020). Such reinforcement schedule effects are a bedrock of operant theory and this article was published in a special “Health, Technology, & Behavior Science” issue of Perspectives on Behavior Science (Q2), the flagship journal of the primary international organization dedicated to operant analysis (ABAI). This group is notoriously fastidious about theoretical considerations and their publication of this work speaks to the rigor of my analytic approaches.

photo by Arek Adeoye


MEDITATION & HABIT FORMATION

Habit formation is another theoretical area that I have aimed to refine via big data analyses. Habits have historically been assessed by self-report instruments consisting of items similar to “Behavior X is something I do automatically.” Such measures have questionable accuracy and do not allow habit initiation to be studied. To address this deficiency, I was awarded a 2020 Kay Family Foundation Data Analytics Grant to investigate the dynamics of how exercise and meditation habits are formed. The mediation data will be provided by Calm, a popular mobile phone application for mindfulness with a five-star rating in both the iPhone and Android application stores and over 2.4 million paid subscribers. This work reflects an exciting new corporate relationship that I have built and plan to continue to develop. I have already had preliminary conversations with Calm’s Scientific Advisory Board about conducting future experiments within their platform. 


DATA CENTRIC METHODOLOGIES

I have also used data-centric methodologies to explore the effects of various character traits on behavioral outcomes. Working with Dr. David Pincus from the Chapman Psychology department, I used techniques from nonlinear dynamical systems analysis to identify markers of behavioral resilience in assessments of both psychopathology and exercise routines. The latter work led to the production of a first-authored manuscript currently under review in the journal Psychology of Sport and Exercise, while the results from the former were published in Nonlinear Dynamics, Psychology, and Life Sciences (Pincus, et al., 2019). In addition to exploring the role of resilience, I have also investigated how exercise behavior is affected by delayed discounting, which is the degree to which an individual is able to forego an immediate reward in favor of a larger, but delayed, future reward. Analytic results (Phillips et al., 2019) demonstrated that intervention effectiveness was influenced by a mismatch between delay discounting preferences and assignment to immediate versus delayed rewards. More recently, I have used longitudinal measures of delayed discounting to demonstrate that immediate rewards modify discounting rates throughout enrollment in an exercise intervention; these results are currently being prepared for publication. 



COLLABORATORS