Use this URL to cite or link to this record in EThOS:
Title: Passive sleep actigraphy : an automated assessment platform for non-contact sleep monitoring
Author: McDowell, Andrew
ISNI:       0000 0004 6350 7992
Awarding Body: Ulster University
Current Institution: Ulster University
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Sleep is a vital physiological process, which is essential for health and wellbeing. Reported sleep problems are, however, commonplace within the general population. Currently, the literature indicates clinically validated, long term sleep monitoring is best achieved using Wrist Based Sleep Actigraphy (WBSA). WBSA, however, is prohibitively expensive to the general population and requires contact with the body to operate, leading to accessibility and user compliance issues. Correspondingly, a literature review found that neither research nor commercial domains have suitably explored the potential for non-contact, actigraphy driven, sleep monitoring. Accordingly, this Thesis presents a series of investigative studies into the development of Passive Sleep Actigraphy (PSA), a bed-based, non-contact method of recording actigraphic data from a bed occupant for the purpose of sleep profiling. The first investigation considered suitable hardware for recording in-bed movement without body contact, where it was discovered that a range of in-bed postural changes could be recorded using a tri-axial accelerometer. A follow up study considered whether PSA data could be used to identify periods of sleep and wake using a WBSA like approach. To evaluate this, 118 hours of validated data was recorded, then used to develop and test a PSA sleep scoring framework. This resulted in metrics of 0.96 sensitivity (sleep detection) and 0.79 specificity (wake detection). A third study focused on optimising the sleep scoring framework, to improve overall performance. This included an evaluation of vector-based features, addressing sleep/wake class imbalance and identifying optimal accelerometer locations. Accordingly, a further 291 hours of validated data was recorded and used to evaluate the revised framework. This presented improvements of up to 16% in specificity, with nominal changes to sensitivity. The final study investigated sleep statistics generated from the PSA data, against those from a number of existing sleep monitoring tools. Observations here indicated that the PSA platform appears to provide comparable measures of sleep to existing approaches. In summary, the work presented throughout this Thesis culminated in a prototype platform for actigraphy driven, non-contact sleep monitoring. Accordingly, PSA addresses several limitations associated with WBSA and offers a novel contribution to objective sleep monitoring through its findings and associated datasets.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available