Meta-Bayes Classifier with Markov Model Applied to the Control of Bioprosthetic Hand

The paper presents an advanced method of recognition of patient’s intention to move of multijoint hand prosthesis during the grasping of objects. In the considered decision problem we assume that each prosthesis operation can be divided into sequence of elementary actions and the patient’s intention means his will to perform a specific elementary action. A characteristic feature of the explored sequential decision problem is the dependence between its phases at particular instants which should be taken into account in the recognition algorithm. The proposed classification method is based on multiclassifier (MC) system working in sequential fashion, dedicated to EMG and MMG biosignals and with dynamic combining mechanism using the Bayes scheme and Markov model of dependences. The performance of proposed MC system with 3 different types of base classifiers was experimentally compared against 3 sequential classifiers for 1—and 2-instant backward dependence using real data concerning the recognition of six types of grasping movements. The results obtained indicate that use of MC system dedicated to the sequential scheme of recognition process, essentially improves performance of patient’s intent classification and that this improvement depends on the type of base classifiers and order of dependence.