PROJECT SUMMARY ABSTRACT
Self-monitoring (SM) is an essential component gold-standard behavioral treatment of obesity (BOT) – the most
prevalent cause of morbidity and mortality in the US. Adherence to dietary self-monitoring is one of the strongest
consistent predictors of treatment outcomes. However, adherence to dietary SM is generally poor due to the
burden of the gold-standard SM approach (i.e., detailed tracking of all food & drink consumed). Discontinuation
of dietary SM is directly correlated with an end to weight loss, and it increases risk of weight regain. Improving
adherence to dietary SM is thus one of the most critical strategies for treating obesity. Alternative SM approaches
have been developed to reduce burden of dietary SM and thereby improve adherence. Prior work shows that
these SM alternatives have the potential to sustain SM adherence and improve weight loss in BOT. However,
there is no scientific consensus in how or when to apply SM alternatives for maximum benefit. To address this
critical gap, we will conduct a micro-randomized trial (MRT) to determine which SM strategy to apply for whom,
and at what point during BOT, to maximize SM adherence and weight loss. MRT involves repeated
randomizations at specific decision points, which enables data-driven optimization of the composition, tailoring,
and timing of behavioral interventions. By using MRT to compare SM strategies over time within individuals and
evaluating factors that influence their efficacy, we can enable broadly-applicable SM recommendations to
improve BOT. We can then use these data to develop algorithms that can repeatedly and automatically adapt
SM recommendations based on each individual’s performance and needs. Our MRT will test the effects of gold-
standard dietary SM and 4 SM alternatives on SM adherence and weight loss during a 24-week BOT. A diverse
sample of adults with overweight/obesity will receive our established, online BOT for 24 weeks. At program start,
and every 2 weeks thereafter, participants will be randomized to one of five SM approaches, for a total of 12
independent randomizations per person. The SM approaches to be tested are: gold-standard full dietary SM of
all foods/drinks and their kcals; reduced-frequency full dietary SM (3 d/wk); SM of dietary lapses only;
smartwatch-based monitoring of energy intake; and SM of body weight only via smart scale. Data from this MRT
will have tremendous scientific and practical impacts; we will evaluate the efficacy of alternative SM approaches
on SM adherence and weight loss as well as how each strategy works across individual differences (e.g., sex),
social determinants of health (e.g., financial resource strain), and time-varying treatment (e.g., week of treatment)
factors. We will use reinforcement learning, a machine learning approach, to create an adaptive SM-selection
algorithm that automatically determines the SM approach most likely to maximize outcomes for a given individual
throughout treatment. This proposal will achieve both scientific and practical public health benefit by using a
sophisticated data-driven approach to understand the factors that influence SM adherence during behavioral
obesity treatment, and provide an algorithm to optimize SM in future clinical and research applications.
