Work with Roselina Arelhi, Frans Coenen, Andrew Hopkinson and Paolo Paoletti.
Session 4: Wearables, Clinical Monitoring, and Sensing Infrastructure; 15:55–16:55, Lecture Theatre 1.
Related research: Sleep wearables.
Back to topOmar Elnaggar 
Abstract
Sleep posture is more varied than a small set of labels such as supine, prone and side-lying can describe. This presentation examines how limb-joint orientations measured with an inertial body sensor network can capture that variation. It brings sensor fusion together with synthetic training examples generated from a single observation of each posture. Studies in simulated and participant-based settings explore what these measurements can distinguish, and where their reliability is limited by orientation drift and similarity between postures. The emphasis is on connecting meaningful biomechanical measurements with methods that can learn from very little labelled data.