B1951
Title: Sleep classification with artificial synthetic imaging data using convolutional neural networks
Authors: Peter Song - University of Michigan (United States) [presenting]
Abstract: A new analytic framework is proposed, Artificial Synthetic Imaging Data (ASID) Workflow, for sleep classification from a wearable device comprising: 1) the creation of ASID from data collected by a non-invasive wearable device that permits real-time multi-modal physiological monitoring of heart rate (HR), 3-axis accelerometer, electrodermal activity, and skin temperature and 2) the use of an image classification supervised learning algorithm, convolutional neural network (CNN), to classify periods of sleep. We compare our ASID Workflow with competing machine/deep learning classification algorithms, including logistic regression, support vector machine, random forest, k-nearest neighbors, and Long Short-Term Memory. The ASID Workflow achieves excellent performance with high mean weighted accuracy and is superior to the Competing Workflow. We explore specifically the influence of data resolution and HR modality on the Workflow's performance in order to achieve the desirable cost-and-effectiveness of data collection. Applying CNN to ASID allows us to capture both temporal and spatial dependency among physiological variables and modalities by using 2D images' topological structure that competing algorithms fail to utilize.
