UC Irvine

The recent developments in signal processing, machine learning, and smart sensing technology allow real-time monitoring and recording of bio-signals, especially the electrocardiogram (ECG) and electroencephalogram (EEG) signals. However, continuous monitoring of those signals is challenging in low-power systems such as wearable devices due to energy and memory constraints. Therefore, in this work, I proposed a novel lightweight single-channel EEG-based method to estimate arousal levels, defined as an individual’s degree of alertness or responsiveness to stimuli at low-power wearable devices for brain activity monitoring. Moreover, a novel resource-efficient template control based Convolutional Neural Network (CNN) architecture is presented for cardiac activity monitoring. To evaluate the proposed methodology performance for brain activity monitoring, I used scalp EEG recorded duringovernight sleep and intra-operative anesthesia with technician-scored hypnogram annotations at the University of California at Irvine Medical Center. And, the well-known PTB diagnostic ECG database is used to assess the proposed methodology’s performance for cardiac activity monitoring. Evaluation of real hardware shows that the proposed methodologies can be implemented for devices with a minimum RAM of 512 KB while maintaining high accuracy with low energy consumption compared to the state-of-the-art works.