Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Jelmer Bot, Dries Testelmans, Jan Aerts (Hasselt University, Hasselt, Belgium; Department of Respiratory Diseases, Leuven University Hospital, Leuven, Belgium;)
Sleep apnea is a common sleep disorder that is often not recognized. This poster presents a topological data analysis (TDA) of clinical data in sleep apnea to discover sub-populations and work towards novel scores that characterize sleep apnea. The purpose of this research was to demonstrate how TDA can be used to explore sleep apnea data. The observation-space and feature-space were modelled using Mapper and STAD, respectively, to explore the data’s structure and discover how clusters differ. We found three subpopulations of patients with a desaturated blood oxygen level in 20 to 50 percent of their total sleep time. Patients in these clusters differed in their BMI, prevalence of obstructive, mixed, and central apnea events, and the number of times they awoke during the night. To conclude, we showed how topological data analysis algorithms can be used to discover sub-populations in sleep apnea data. Given that these algorithms only require the distance between data-points to work, they can be used on a wide range of data.