We developed three separate models that provide hour-by-hour probabilities of a future lapse back to alcohol use with increasing temporal precision (i.e., lapses in the next week, next day, and next hour). Model features were based on raw scores and longitudinal change in theoretically implicated risk factors collected through ecological momentary assessment (EMA). Participants (N=151; 51% male; mean age = 41; 87% White, 97% Non-Hispanic) in early recovery (1–8 weeks of abstinence) from alcohol use disorder provided 4x daily EMA for up to three months. We used grouped, nested cross-validation, with 1 repeat of 10-fold cross-validation for the inner loop and 3 repeats of 10-fold cross-validation for the outer loop to train models, select best models, and evaluate those best models on auROC. Models yielded median areas under the receiver operating curves (auROCs) of .90, .91, and .94 in the 30 held-out test sets for week, day, and hour level models, respectively. Some feature categories consistently emerged as being globally important to lapse prediction across our week, day, and hour level models (i.e., past use, future efficacy). However, most of the more punctuate, time varying constructs (e.g., craving, past stressful events, arousal) appear to have greater impact within the next hour prediction model. This research represents an important step toward the development of a smart (machine learning guided) sensing system that can both identify periods of peak lapse risk and recommend specific supports to address factors contributing to this risk. General scientific summary: This study suggests that densely sampled self-report data can be used to predict lapses back to alcohol use with varying degrees of temporal precision. Additionally, the contextual features contributing to risk of lapse may offer important insight for treatment matching through a digital therapeutic.