Time-frequency analysis of accelerometry data for seizure detection

This article presents an algorithm based on a short-time Fourier transform for reliable detection of epileptic seizures measured with three-dimensional (3D) accelerometry. The objective of the described work is to provide basic technical information to create useful alarm models for epileptic seizure detection using a mobile phone. The presented material is based on experimental measurements. Finally, the possibility of increasing smartphone detection capability by attaching a triaxial piezoelectric accelerometer to the patient’s wrist is suggested.