Handwriting-Based Gender Classification Using Robotic and Machine Learning Models

Abstract Handwriting analysis provides insights into motor control and cognitive processes, with potential differences arising from biological gender and neurological conditions such as Parkinson’s disease (PD). Investigating these differences can lead to improved understanding of motor and cognitive functions. This study introduces a novel methodology that integrates robotic features to estimate gender from handwriting. Kinematic and dynamic features are estimated by simulating handwriting with a robotic model. Linear predictive coding (LPC) and singular spectrum analysis (SSA) are applied to the kinematic and dynamic sequences. Machine learning algorithms are used to classify handwriting as male or female. Handwriting samples from healthy individuals (BiosecurID database) and PD patients (PaHaW dataset) were analyzed. The proposed method demonstrates state-of-the-art performance in gender classification, revealing significant differences between healthy and unhealthy individuals. The robotic-based approach successfully mimics arm movements during writing, highlighting distinct motor patterns associated with gender and health status. This research advances the understanding of gender-based differences in motor and cognitive function, particularly in populations with neurological conditions. The integration of robotic features and machine learning provides a promising pathway for future investigations in handwriting analysis, gender classification, and neurodegenerative disease diagnosis.