A Non-EEG Biosignals Dataset for Assessment and Visualization of Neurological Status

Abstract

Neurological assessment can be used to monitor a person's neurological status. In this paper, we report collection and analysis of a multimodal dataset of Non-EEG physiological signals available in the public domain. We have found this signal set useful for inferring the neurological status of individuals. The data was collected using non-invasive wrist worn biosensors and consists of electrodermal activity (EDA), temperature, acceleration, heart rate (HR), and arterial oxygen level (SpO2). We applied an efficient non-linear dimension reduction technique to visualize the biosignals in a low dimension feature space. We could cluster the four neurological statuses using an unsupervised Gaussian Mixture Model. The experimental results show that our unsupervised method can accurately separate different neurological statuses with an accuracy of greater than 84%.