This study investigates the use of entropy metrics to enhance traditional analysis of grasping movements. The authors examined whether signal unpredictability measures—Approximate Entropy (ApEn) and Sample Entropy (SampEn)—can reveal motor control aspects during reach-to-grasp actions. They aimed to determine if entropy could infer the movement’s underlying intention, distinguishing social (SOC) versus individual (IND) goals. Wrist acceleration data were collected via wearable Inertial Measurement Units (IMU) from 12 older adults performing repetitive reach-to-grasp tasks in two conditions: IND (acting alone) and SOC (passing an object to an experimenter). The hypothesis predicted that social actions involve tighter motor control, resulting in lower entropy. ApEn and SampEn were computed using different parameter sets (embedding dimension m = 2, 3; similarity criterion r = 0.1, 0.2, 0.3). Mixed-model repeated measures ANOVAs and linear mixed-effects models assessed differences between conditions and parameters. Results showed significant effects of condition, m, and r, including m × r interactions. Both ApEn and SampEn yielded lower entropy in the SOC condition, confirming the hypothesis. ApEn showed greater robustness and consistency across parameter settings and preprocessing steps compared to SampEn. This study is the first systematic comparison of ApEn and SampEn for action-related time series in social versus non-social contexts. Findings highlight the impact of social context on motor control. ApEn, especially with lower m (2) and r (0.1–0.2), is recommended for future studies due to its reliability and sensitivity in distinguishing movement intention.
Mancioppi et al. (Thu,) studied this question.