Toward filtering of athetoid motion with neural networks

People with athetoid cerebral palsy (CP) have difficulty using computers due to unintentional involuntary movements in the upper extremities. A neural network-based system has been developed to cancel the undesired motion, and speed up the movements and accuracy in target acquisition and path tracking tasks while using an isometric joystick (IJ). Nonlinear filtering algorithms were created with neural networks using nonlinear models to help people with athetoid CP to access the computer. This paper presents unfiltered test data that have been collected from patients, and describes the planned filtering approach.