Unobtrusive sleep monitoring


Speaker:   Dr. Xi Long

Time:       15:00-16:00, Jan. 9

Location:  SIST 1A-200

Host:       Prof. Lin Xu



In humans, sleep is important. For adults, sleep is a state of reversible disconnection from the environment and plays an exceptionally essential role in maintaining internal homeostasis, memory consolidation, energy conservation, and cognitive and behavioral performance. However, problems in sleeping are widely prevalent globally with increasing sleep complaints due to the influence of artificial environments where lighting, heat, and food are available at any moment, leading to epidemic levels of sleep disturbances and disorders. For infants or children, in particular neonates under intensive care, sleep is paramount for the brain development, during which neural connections are formed and the development of brain regions is triggered. All these factors are highly correlated with sleep stages.

The current standard of sleep monitoring relies on polysomnography (PSG) or behavioral observation (for infants) that is able to assess sleep quality and potentially detect pathological sleep events. However, PSG has primary problems, requiring costly facilities, many sleep disturbing devices with electrodes and wires attached to the body, and time-consuming manual effort from trained sleep clinicians to score sleep based on multi-channel signals or visual observation. These problems disable the monitoring to be continuous and efficient. In neonates, more importantly, the use of adhesive electrodes can cause damage to their fragile skin, making them more prone to infections. Therefore, unobtrusive or even non-contact methods are urged.

Cardiorespiratory and movement activities have been shown to associate with sleep stages because of the regulation of autonomic (sympathetic and vagal) nervous system. The most pronounced advantage is that these physiological activities can be reliably acquired using unobtrusive, wearable, or non-contact sensors due to the rapid development of biosignal sensing technologies in the past decade, such as photoplethysmography, ballistocardiography, capacitive sensing, camera, and Doppler radar. Using these technologies would potentially solve the problems of PSG and the combination of signal processing and artificial intelligence algorithms is the key enabler for automatic sleep monitoring.

This talk will introduce our previous work and current advancement of monitoring and classifying sleep (stages) based on cardiac, respiratory, and body movement data.


Dr. Long obtained his BEng with honor in electronic information engineering from Zhejiang University, China, and MSc in electrical engineering from the Eindhoven University of Technology (TU/e), the Netherlands, in 2006 and 2009, respectively. He worked at Philips Research as a researcher from 2008 to 2009 and at Tencent (China) as a project manager and data mining engineer from 2009 to 2011. He then went on to pursue his PhD at TU/e and Philips Research, on signal processing and machine learning in unobtrusive sleep monitoring, which he completed with cum laude distinction in 2015.

Currently, Dr. Long is a Senior Scientist at Philips Research, and an Assistant Professor in the Signal Processing Systems (SPS) Group, Department of Electrical Engineering, TU/e, the Netherlands. He advises, coordinates and participates in many in-depth collaborative projects with academic and clinical partners, for example, Maxima Medical Center, Kempenhaeghe Sleep Medicine and Epilepsy Center, University Medical Center Utrecht, Academic Medical Center Amsterdam, and Donders Institute (the Netherlands), KU Leuven (Belgium), RWTH Aachen (Germany), Imperial College London (UK), SUNY Buffalo (US), and Fudan University (China).

Dr. Longs expertise is in signal processing, time series analysis, machine learning, deep learning, data analytics, network analysis and modeling. His research interests include engineering for biomedical/healthcare applications such as unobtrusive sensing, patient monitoring, sleep, physical activity, perinatal and pregnancy monitoring, electrophysiological and psycho-physiological analyses, cardiorespiratory dynamics, epilepsy, and brain activity. In addition, Dr. Long has published over 80 scientific articles and reports, generated more than 25 Philips inventions or patent filings and supervised more than 35 PhD or MSc students. He served as a program committee member of several international conferences such as IEEE DSP, IEEE BSN, and CVPM/CVPR. He is a reviewer of more than 30 prestigious international journals in his research areas including Nat Commun, IEEE Trans Biomed Eng, IEEE Trans Indus Inform, IEEE Biomed Circuits Syst, IEEE J Biomed Health Inform, Sci Rep, J Neurosci Methods, J Neural Eng, etc.

SIST-Seminar 18234