Ä¢¹½Ö±²¥

Sleep issues have a negative impact on global population health. Zhou's dissertation focuses on deep learning-based methods for automatic sleep scoring using a single-channel electroencephalogram (EEG). The proposed methods could achieve promising performance in sleep scoring on different datasets.

Sleep issues are on the rise and have a negative impact on global population health, particularly during the COVID-19 outbreak. For instance, approximately 10-30% of the population worldwide exhibit insomnia symptoms. Accurate sleep disorders diagnosis and sleep quality evaluation will significantly lessen suffering and raise the standard of living.

Traditional manual sleep scoring is time-consuming and tedious, so it is essential to develop automatic sleep scoring methods to fulfill the growing unmet demands for sleep research.

Multi-modal polysomnography (PSG) recordings are considered the golden tool in clinics. However, it is still challenging to realize reliable results using as few channel signals as possible. In this dissertation, the focus is on deep learning-based methods for automatic sleep scoring with single-channel electroencephalogram (EEG).

The proposed methods could achieve promising performance in sleep scoring on different datasets. The single-channel EEG scheme is more comfortable to wear for practitioners and is more suitable to utilize in portable sleep monitoring devices.

Ultimately, we expect this thesis to contribute to the practical application of automatic sleep scorning methods and service personal well-being promotion in the future.

M.Sc. Dongdong Zhou defends his doctoral dissertation in Software and Communications Engineering "Automatic sleep stage classification based on single-channel EEG". Opponent Professor Chengyu Liu (Southeast University, China) and Custos Professor Lauri Kettunen (Ä¢¹½Ö±²¥). The doctoral dissertation is held in English.

Audience can follow the dissertation in the lecture hall Agora Auditorium 2, or online. Link to the Zoom Webinar event (Zoom application or Google Chrome web browser recommended): 

Permanent link to the dissertation in the JYX publications archive: