Wearable, lightweight single-channel EEG for the detection of urinary urgency and voiding intentions.

BACKGROUND

This preliminary study explores the potential of wearable single-channel electroencephalography (EEG) as a non-invasive biomarker for identifying urinary urgency and voiding intention, supported by the development of a lightweight, real-time processing algorithm.

MATERIALS AND

METHODS

A total of 20 individuals were recruited, and the EEG signals were recorded using a single-channel electrode placed on the frontal cortex. A lightweight convolutional neural network architecture based on a modified DenseNet was proposed, with only 75,194 trainable parameters. The performance was evaluated based on accuracy, precision, recall, and F1-score.

RESULTS

The lightweight proposed model achieved an accuracy of 0.8457, precision of 0.8540, recall of 0.9100, and F1-score of 0.8811 in classifying EEG signals for urgency. A 20-second epoch length was found optimal, balancing temporal richness and sample availability. Simple artifact detection methods, particularly slope thresholding, significantly improved performance. Compared with other lightweight architectures, the proposed model demonstrated superior accuracy and efficiency, making it well-suited for real-time and embedded EEG applications.

CONCLUSION

This study demonstrates the feasibility of using wearable single-channel EEG as a non-invasive biomarker for detecting urinary urgency and voiding intention. These findings may facilitate the development of personalized and adaptive intelligent rehabilitation systems for managing lower urinary tract dysfunctions.