Linear Mixed Model of ROC-AUC for Smoking Cessation JITAI Predictions

Corinna Leppin published a dataset on figshare on 2026-05-14 containing results from a study optimizing ecological momentary assessment (EMA) for smoking cessation just-in-time adaptive interventions (JITAIs). The 5.5 KB XLS file presents linear mixed model results evaluating random forest algorithm performance for predicting smoking lapses and cravings based on data from 37 participants. Performance metrics include F1-scores and ROC-AUC values across different EMA prompt frequencies, predictor counts, and training data sources.

Use Cases

• Evaluating the impact of EMA prompt frequency on machine learning prediction performance for smoking lapses and cravings.
• Comparing the performance of lapse prediction versus craving prediction models using ROC-AUC and F1-score metrics.
• Assessing the effect of feature reduction and participant-specific training data on model performance in a JITAI context.

Strengths

• Contains specific performance metrics (e.g., Median ROC-AUC for Lapses 0.659, for Cravings 0.628) derived from a study with 37 participants.
• Explicitly compares results across multiple experimental configurations, including 16, 4, and 3 EMA prompts per day.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

figshare

Collection Method

Data likely contains aggregated results from linear mixed models analyzing random forest algorithm performance on ecological momentary assessment (EMA) data.

Freshness

Last updated 2026-05-14 17:33:22; freshness should be verified.