Development of a Predictive Classification Model Using Artificial Neural Networks Based on Selected Biomotor Abilities in Foil Fencing Students

Abstract

This study aimed to develop a predictive classification model based on Artificial Neural Networks (ANNs) to identify skill performance levels in foil fencing using selected biomotor abilities. The researchers adopted a descriptive survey approach on a purposive sample of (70) students from the College of Physical Education and Sports Sciences at the University of Karbala during the academic year (2025–2026). Participants underwent a standardized battery of physical and motor tests designed to assess biomotor abilities associated with fencing performance. The Least Absolute Shrinkage and Selection Operator (LASSO) technique was employed to identify the most significant predictive variables. Subsequently, a Multilayer Perceptron (MLP) neural network model with a single hidden layer was developed to classify skill performance levels. The results revealed statistically significant differences in biomotor abilities among participants, indicating their substantial contribution to predicting skill performance in foil fencing. The ANN model demonstrated acceptable classification performance, achieving prediction accuracy rates ranging from 64% to 71%. The model also showed greater efficiency in distinguishing extreme performance levels compared with intermediate levels. These findings highlight the effectiveness of integrating machine learning techniques, particularly artificial neural networks, in enhancing the objectivity of sports talent identification and improving evidence-based training program design. The study recommends increasing sample sizes, incorporating psycho-cognitive variables, and employing deeper neural network architectures in future research to improve predictive accuracy and model generalizability