Developing AI-Powered Personalized Training Systems for Traditional Sports to Enhance Athlete Engagement and Performance

This study presents a Dynamic Multi-Sensor Fusion Classifier (DMSFC) for biometric-driven athlete monitoring by integrating heterogeneous modalities including ECG signals, fingerprint patterns, and facial features. The proposed framework combines multi-instance learning with ensemble-based decision-level fusion to enhance classification robustness under cross-modal variability. A Hybrid Mutual Information Entropy Criterion (HMIEC)-based feature selection mechanism is employed to identify discriminative features while suppressing inter-sensor redundancy. System performance is evaluated using k-fold cross-validation with metrics including accuracy, sensitivity, specificity, and area under the ROC curve (AUC). Experimental results demonstrate that DMSFC achieves superior classification performance compared with standalone classifiers, indicating improved stability, reliability, and discrimination capability. The findings highlight the effectiveness of decision-level multimodal fusion for real-time biometric monitoring and athlete performance assessment applications.