- This event has passed.
Insigneo Seminar: I dream of chainsaws – methods for the automatic assessment of sleep-disordered breathing
Friday, 24 May, 1:00 pm - 2:00 pm BST
We are delighted to announce that on 24 May 2019 Richard Wiffen, Director of Passion for Life Healthcare will give a joint seminar with Professor Guy Brown, Head of the Department of Computer Science, on the subject of ‘I dream of chainsaws – methods for the automatic assessment of sleep-disordered breathing’
Sleep-disordered breathing (SDB) results from the collapse of the upper airway during sleep. Its most prevalent forms are snoring, caused by a partial collapse, and obstructive sleep apnoea (OSA) which is due to a complete collapse. Polysomnography is the current gold-standard test for diagnosing SDB, but it is obtrusive, time-consuming and relatively expensive. There is therefore a need for low-cost, reliable home monitoring and diagnosis of SDB. The presenters have been collaborating on the development of a smartphone-based solution for assessment of SDB, via projects funded by Innovate UK.
In the first part of this talk, recent evidence of the effect of anatomy and gender on the acoustic characteristics of SDB will be reviewed. We then describe two studies which have developed computer-based methods for assessment of SDB. The first of these recorded a substantial corpus of SDB sounds, and developed machine-learning methods for classification of acoustic events (snore, breath, silence etc.). Finally, we describe methods for ensuring that SDB event classification is robust to the sounds made by a bed partner, an important consideration if this technology is to be used in home environments.
Professor Guy Brown
Professor Guy Brown is Head of the Department of Computer Science, University of Sheffield. His main research interest is Computational Auditory Scene Analysis (CASA), which aims to build machine systems that mimic the ability of human listeners to segregate complex mixtures of sound. He also has interests in noise-robust and reverberation-robust automatic speech recognition, models of auditory function in normal and impaired hearing, binaural modelling and the phonetics of overlapping speech.