Core Concepts
Time2Stop is an intelligent, adaptive, and explainable JITAI system that significantly reduces smartphone overuse through personalized interventions and user feedback.
Abstract
Time2Stop developed by a team of researchers from various universities.
Conducted an 8-week field experiment with 71 participants to evaluate the effectiveness of adaptive and explainable aspects.
Results show adaptive models outperform baseline methods in intervention accuracy and receptivity.
Incorporating explanations enhances effectiveness and reduces app visit frequency.
Participants preferred adaptive interventions and rated the system highly.
Time2Stop integrates ML for optimal intervention timings, explanations, and user feedback.
Implemented on Android OS with a cloud-based ML pipeline.
Experiment design included four intervention types: Control, Personalized, Adaptive-wo-Exp, and Adaptive-w-Exp.
Micro-randomized trials used to evaluate intervention types.
Evaluation metrics included intervention accuracy, receptivity, app usage duration, and visit frequency.
Participants' feedback and preferences were collected through questionnaires and interviews.
Stats
"Our results indicate that our adaptive models significantly outperform the baseline methods on intervention accuracy (>32.8% relatively) and receptivity (>8.0%)."
"Moreover, Time2Stop significantly reduces overuse, decreasing app visit frequency by 7.0∼8.9%."
Quotes
"Participants preferred the adaptive interventions and rated the system highly on intervention accuracy, effectiveness, and level of trust."