Collection and exploratory analysis of attitude sensor data in an epilepsy monitoring unit.

The aim of this paper is to present the collection of attitude sensor data from an epilepsy monitoring unit and the results of standard exploration using principal component analysis. The collection of data from attitude sensors positioned on three limbs of epileptic patients at their bedside is described. The analysis of the data focuses, on one hand, on motor features extraction from attitude sensor data and on the other hand, on visual segmentation of seizures into events corresponding to motor manifestations classes by an expert. Principal component analysis is then realized over these features and groups of data are localized according to the expert classification. This exploration indicates a possible discrimination between these motor manifestation classes.