Adaptive band-pass filter (ABPF) for tremor extraction from inertial sensor data

Cancelling pathological tremor in everyday living activities may be possible with functional electrical stimulation (FES). One such feasible FES system with feedback from inertial sensors would rely on tremor estimates in real time. We present an adaptive band-pass filter (ABPF) that estimates tremor from volitional hand movement with zero-phase lag. The proposed algorithm is simple and easy to implement. Performance of the ABPF is compared to one popular well-established method for tremor extraction (weighted-frequency Fourier linear combiner, WFLC) using both synthetic data and data from inertial sensors, recorded in tremor patients during "finger to nose" task execution. The results were comparable, favoring ability of ABPF for faster adaptation, higher accuracy and robustness.