Research And Application On Several Classification Methods Of Body Characteristic Signals

Focus on the problems of ECG data's computer-aided diagnosis, firstly, four classification models, i.e. improved independent component analysis and supporting vector machine model, the linear prediction and the principal component analysis model, certainty factor model based on expert system, Hidden Markov model, have been introduced in this paper. Then, the MIT-BIH database has been selected to test and improve these models considering the special natures of ECG data, and the advantages and shortcomings of the four models have been discussed and analyzed.Subsequently, the practical twelve-lead ECG data have been discussed. This study proposed various improvement ideas for the ultimate aim of practical applications. Through the analysis and discussion, the improved independent component analysis and support vector machine model have been chose to be the transplantation aim, and some valuable results have been obtained by a few methods such as feature selection and feature integration.Of course, this paper is just a milestone results for application, we must keep going on transplantation, feedback and improvement to meet the actual requirement.