I-CRAV: Intervention where and when you need it: smartphone-based Food-Approach-Avoidance trainings during high CRAVing times (I-CRAV)

Wider research context: Regulating food intake is challenging, partly due to people's inclination to approach tasty, calorie-dense foods. Such approach tendencies may interact with food craving to override deliberate intentions, resulting in unhealthy food consumption. To curb this, recent laboratory studies have shown that repeatedly pairing images of unhealthy foods with avoidance reactions in so-called Approach-Avoidance interventions (AAI) can reduce unhealthy food intake. However, it remains untested whether the efficacy of AAIs can be increased when delivered during high-risk times via mobile devices.
Objectives: The proposed study series tests the effectiveness of a smartphone-based mobile AAI training to reduce the risk of dietary lapses. By examining the role of food craving before AAI performance, this project contributes to theoretical debates about AAI effects and helps to determine ideal intervention timing when delivered in everyday life. In a further step we use machine learning techniques to dynamically predict individual states of high lapse risk and deliver AAI before these states arise in a preventive approach.
Approach: In a 2-by-2 full factorial design (total n=400), Study 1 will use a craving-induction procedure (vs. low-craving induction) before testing the effects of AAI (vs. sham) on changes in desire to consume presented foods, food liking, and food choice in the laboratory. In Study 2 (total n=150), we use data obtained through ecological momentary assessment (EMA) to identify current cravings and deliver AAI when such states are detected to be either high or low. Instead of detecting current cravings, interventions in Study 3 (n=150) use predictions of upcoming high craving states that are derived from data of a two-week EMA phase. Thus, in the three-week intervention phase, idiosyncratic prediction algorithms deliver AAIs before epochs with the likely occurrence of strong cravings in order to reduce the probability of subsequent dietary lapses. Such algorithm-based just-in-time intervention will be compared against random intervention times