Researchers searched PsycInfo, MEDLINE, and Scopus for studies using ecological momentary assessment (EMA) or ecological momentary intervention (EMI) methods to measure or address substance use among youth, defined as a mean or median sample age of 25 or younger, in clinical settings. Eligible settings included emergency departments, general medical care, mental health services, and substance use treatment programs.

From 12,093 titles and abstracts screened and 2,157 full texts reviewed, the review ultimately included 43 published studies representing 24 unique monitoring or intervention protocols. Fifteen protocols were purely observational, tracking substance use patterns without intervening, and nine tested an active real-time intervention.

Across the 24 protocols, alcohol was the most frequently studied substance (75%), followed closely by cannabis (67%) and nicotine or tobacco (33%). Many protocols tracked more than one substance simultaneously, reflecting the reality that polysubstance use is common among the youth populations studied.

The typical EMA protocol asked participants to complete brief check-ins for 14 days, with four prompts delivered per day, a burden low enough to sustain the two weeks needed to capture meaningful patterns. Three protocols went further, adding passive biosensing, wearable or phone-based sensors that track physiological or behavioral signals, without requiring the participant to actively respond.

The included studies captured 270 distinct momentary antecedents (factors that came before substance use) and 17 momentary consequences (effects that followed it), organized by the reviewers into six domains: substance use itself, affect and cognition, environmental context, social context, movement or physical activity factors, and how consistently participants completed their EMA prompts.

Three factors showed the most consistent association with substance use across studies: negative affect (feeling stressed, anxious, or low), situational context such as evenings, weekends, or being around friends, and prior substance use earlier in the same monitoring period. The review’s authors describe the overall findings as heterogeneous, meaning the strength and even the direction of some associations varied across studies.

Beyond monitoring alone, the review identified 13 evaluated interventions that layered a real-time response, feedback about a person’s own reported use, goal-setting prompts, or coping-skill support, on top of ecological momentary monitoring. These interventions were associated with reductions in substance use-related harms in the included studies.

The review also notes that passive sensing approaches, the wearable or phone-based tools that do not require active participant input, demonstrated strong predictive performance in the studies that used them, a signal that these tools may eventually reduce the burden of frequent active check-ins.

This review was posted to Research Square, a preprint server, on July 30, 2026, and carries PubMed identifier 42619989 with the article type listed as “Journal Article, Preprint.” It has not yet completed formal peer review or been accepted by a journal at the time of this writing.

Readers, and clinicians in particular, should weigh this review’s findings with that status clearly in mind. Preprints allow findings to circulate faster, but they have not yet been vetted by independent expert reviewers, and the specific conclusions, methods, and data reported here could change before, or if, this work is formally published.

Interest in digital phenotyping and real-time monitoring for substance use has grown alongside the broader adoption of smartphones and wearable sensors in clinical research, offering a way to study behavior as it happens rather than reconstructed from memory during a clinic visit or research interview.

This review adds to that broader movement by focusing specifically on youth, a population where substance use screening in traditional clinical encounters is known to be imperfect, and where the treatment settings, emergency care, mental health services, and substance use programs, represent moments when clinicians have a real opportunity to intervene early.

I welcome any serious attempt to understand cannabis and other substance use in young people as it actually happens, rather than through the imperfect lens of a single clinic visit and a patient’s memory of the weeks before. The pattern this review surfaces, that negative mood, specific situational triggers, and prior use itself are the most consistent predictors of subsequent use, matches a great deal of what I hear directly from young patients and their families.

That said, I want to be plain about where this evidence currently stands. This is a preprint. It has not yet been peer-reviewed, and I am treating its specific findings as an early, exploratory signal rather than settled science. I am genuinely encouraged by the direction, real-time tools that can catch a risk moment and respond to it, but I would not recommend a specific app, protocol, or intervention to a patient or family based on this review alone. I will be watching for the peer-reviewed version of this work, and for the larger, better-powered trials the field still needs, before drawing firmer clinical conclusions.

A Four-Step Reading Frame