Evaluation of Grasp-and-Extend Hand Dynamics and Intelligent Modeling of Grasp Hand Dynamics

Abstract: This master thesis looks to find dynamics in hand movements to further be used in stroke rehabilitation research. To be able to provide feedback during rehabilitation could help with movement recovery for a stroke patient for whom these movements have been damaged. Data was recorded from both right and left hands of the participants. The data was evaluated, both between different participants and different hands on the same participant. Different movements were separated to be examined individually. Both classification and prediction was performed and Radial Basis Function Neural Networks were used for classification as well as for prediction. The data used in the prediction training was the data recorded while the hand was closing and the data used in the classification training was the data recorded while the hand was closing into a grip and relaxing again. The classification was able to find classes in the data successfully. The prediction proved more problematic. However, the results show that there are definite improvements that could be done to improve the networks performances while predicting the movement and that the radial basis function neural network still is interesting to consider in future research on the subject.