Behavioral motion patterns to predict child mobile device use: Evaluating model performance across a standardized screen time protocol - Summary - MDSpire
To assess whether machine-learning models using mobile-device motion-sensor data can distinguish children from their parents during standardized game play, including when tested on games excluded from training.
Approach:
Study design: The study evaluated motion-sensor-based behavioral biometrics under a standardized screen-time protocol, addressing variation in game choice that limited earlier work.
Participants: Thirty-seven parent–child dyads were recruited in one southeastern metropolitan city between April and September 2025. Children were required to be 5–12 years old; the provided context does not include further eligibility or protocol details.
Model comparisons: The study planned to compare performance on games represented in training with performance on unseen games, and to compare training conditions that included or excluded the child’s parent.
Interpretation:
The supplied excerpt describes the study rationale and evaluation aims but ends before reporting model-performance results.
Limitations:
The supplied excerpt ends before the participant eligibility description is complete and does not include study results.
by Olivia L. Finnegan, Bridget Armstrong, Elizabeth L. Adams, Sarah Burkart, Michael W. Beets, James W. White, III, Joshua Culverhouse, Srihari Nelakuditi, Zifei Zhong, Hongpeng Yang, Keagan P. Kiely, Rahul Ghosal, Robert G. Weaver
