Testing a Machine Learning–Based Adaptive Motivational System for Socioeconomically Disadvantaged Smokers (Adapt2Quit): Protocol for a Randomized Controlled Trial
Bibliographic record
Abstract
BACKGROUND: Individuals who are socioeconomically disadvantaged have high smoking rates and face barriers to participating in smoking cessation interventions. Computer-tailored health communication, which is focused on finding the most relevant messages for an individual, has been shown to promote behavior change. We developed a machine learning approach (the Adapt2Quit recommender system), and our pilot work demonstrated the potential to increase message relevance and smoking cessation effectiveness among individuals who are socioeconomically disadvantaged. OBJECTIVE: This study protocol describes our randomized controlled trial to test whether the Adapt2Quit recommender system will increase smoking cessation among individuals from socioeconomically disadvantaged backgrounds who smoke. METHODS: Individuals from socioeconomically disadvantaged backgrounds who smoke were identified based on insurance tied to low income or from clinical settings (eg, community health centers) that provide care for low-income patients. They received text messages from the Adapt2Quit recommender system for 6 months. Participants received daily text messages for the first 30 days and every 14 days until the end of the study. Intervention participants also received biweekly texting facilitation messages, that is, text messages asking participants to respond (yes or no) if they were interested in being referred to the quitline. Interested participants were then actively referred to the quitline by study staff. Intervention participants also received biweekly text messages assessing their current smoking status. Control participants did not receive the recommender messages but received the biweekly texting facilitation and smoking status assessment messages. Our primary outcome is the 7-day point-prevalence smoking cessation at 6 months, verified by carbon monoxide testing. We will use an inverse probability weighting approach to test our primary outcome. This involves using a logistic regression model to predict nonmissingness, calculating the inverse probability of nonmissingness, and using it as a weight in a logistic regression model to compare cessation rates between the two groups. RESULTS: The Adapt2Quit study was funded in April 2020 and is still ongoing. We have completed the recruitment of individuals (N=757 participants). The 6-month follow-up of all participants was completed in November 2024. The sample consists of 64% (486/757) female participants, 35% (265/757) Black or African American individuals, 51.1% (387/757) White individuals, and 16% (121/757) Hispanic or Latino individuals. In total, 52.6% (398/757) of participants reported having a high school education or being a high school graduate; 70% (529/757) smoked their first cigarette within 30 minutes of waking, and half (379/757, 50%) had stopped smoking for at least one day in the past year. Moreover, 16.6% (126/757) had called the quitline before study participation. CONCLUSIONS: We have recruited a diverse sample of individuals who are socioeconomically disadvantaged and designed a rigorous protocol to evaluate the Adapt2Quit recommender system. Future papers will present our main analysis of the trial. TRIAL REGISTRATION: ClinicalTrials.gov NCT04720625; https://clinicaltrials.gov/study/NCT04720625. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/63693.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.027 | 0.028 |
| Meta-epidemiology (narrow) | 0.006 | 0.003 |
| Meta-epidemiology (broad) | 0.009 | 0.004 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.004 | 0.003 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.007 | 0.009 |
| Insufficient payload (model declined to judge) | 0.079 | 0.012 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".