SensoriMind-Trans Net: EEG and sensorimotor-driven transformer for athlete potential evaluation.

Area of Science:

• Sports Science
• Neuroscience
• Biomedical Engineering

Background:

• Traditional athlete potential evaluation methods analyze electroencephalogram (EEG) and somatosensory data separately.
• Existing shallow models fail to capture complex temporal dependencies and cross-modal interactions, limiting comprehensive athlete assessment.
• There's a need for advanced models that integrate multimodal data for a holistic view of athletic capabilities.

Purpose of the Study:

• To propose a novel Transformer-based model, SensoriMind-Trans Net, for integrating EEG and somatosensory data.
• To enhance the comprehensive evaluation of athletes' cognitive and motor abilities through multimodal data fusion.
• To improve the accuracy and robustness of athlete potential assessment.

Main Methods:

• Developed a Transformer-based architecture (SensoriMind-Trans Net) integrating EEG signals and somatosensory data.
• Employed a multi-layer Transformer network to capture temporal dependencies within EEG signals.
• Utilized a somatosensory data feature extractor and cross-modal attention alignment mechanism for enhanced multimodal analysis.

Main Results:

• The proposed SensoriMind-Trans Net demonstrated superior performance compared to existing state-of-the-art (SOTA) models.
• Achieved higher accuracy in athlete potential evaluation across four public datasets.
• Showcased improved inference time and computational efficiency.

Conclusions:

• SensoriMind-Trans Net offers a significant advancement in athlete data analysis by effectively integrating multimodal information.
• The model provides a robust and accurate solution for athlete potential evaluation.
• This work has broad applicability and implications for future multimodal sports performance assessment.