Advanced design and classification of wearable near-infrared spectroscopy device using temporal channel reconfiguration multi-graph convolutional neural networks for motor activity.

In this paper, advanced design and classification of wearable near-infrared spectroscopy device using temporal channel reconfiguration multi-graph convolutional neural networks for motor activity (WNISD-TRMCNN) are proposed. Input data is collected from real-time fNIRS data. The input data are pre-processed using event-triggered consensus Kalman filtering (ETCKF) to remove motion artefacts. Then, the pre-processed data is fed to TRMCNN for classifying wearable NIRS as oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR). To enhance classification, Young's double slit experiment optimization algorithm (YDSEOA) is applied. Performance metrics such as accuracy, precision, AUC, and processing time demonstrate the proposed method's superiority over existing techniques.