Feasibility of channel reduction in real-time estimation of perceptual nociceptive threshold based on EEG measurements

This study aims to advance real-time nociceptive threshold tracking by evaluating perceptual thresholds through intra-epidermal electrical stimulation and automatic classification. Building on a previous study implementing a 32-EEG electrode setup, this research focuses on reducing the number of electrodes needed by exploring different neural network architectures and performance strategies, including hyperparameter tuning, transfer learning, and ensemble models, using data from just eight electrodes. Although the experiments revealed challenges such as high variability and poor correlation between threshold estimates from different tasks, the study provides valuable insights for future research, emphasizing the need for improved electrode configurations and recording methods to optimize performance with fewer electrodes.