Abstract This study presents and validates a dataset designed to evaluate the accuracy of a smartphone application for measuring vertical jump time. A total of 550 trials were recorded, with jump flight time simultaneously measured by a smartphone (Android) and a reference wearable accelerometer (BioPlux). Two predictive models, Least Squares (LSQ) and Multilayer Perceptron (MLP), were trained to estimate BioPlux flight time from smartphone readings. The LSQ model achieved a mean error of 0.43% and a mean absolute error of 5.32%, while the MLP model obtained 1.2% and 5.36%, respectively. Both models showed low average percentage error relative to the reference system. This work provides a robust dataset and modeling framework for evaluating low-cost, mobile-based movement assessment tools, with applications in neurology, rehabilitation, and sports biomechanics.
Radid et al. (Wed,) studied this question.