Using a Quartile-based Data Transformation for Pain Intensity Classification based on the SenseEmotion Database

Abstract

The SenseEmotion Database was collected at Ulm University for research purposes in the field of e-health. The participants of the SenseEmotion data acquisition experiments were healthy subjects exposed to three personalised levels of artificially induced pain under strictly controlled conditions. Our study focuses on the recordings from the physiological sensors, such as electrocardiography and the skin conductance. Based on that part of the data set, we propose using an unsupervised quartile-based data transformation approach, which removes outlier values for better nearest neighbour classification.