Research On Human Body Component Prediction Model Based On Support Vector Machine

Bioelectrical impedance analysis(BIA)has the advantages of safety,simplicity and high efficiency,and is widely used in the modeling and analysis of human body composition.In order to solve the problem that human body composition modeling has many problems such as parameter redundancy and parameter nonlinearity,this paper mainly studies the human body impedance model,the human body characteristic parameter selection algorithm and the human body composition model prediction method.Innovative work is as follows.Firstly,According to the shortage that traditional five-segments model of human body impedance can't satisfy the needs of figuring out human trunk subdivision impedance and the exiting asymmetric eight-segments impedance model has no perfect analysis method,the asymmetric eight-segment human impedance measurement model and its solving method are put forward,which uses the five-segment impedance model to complement the impedance relation of the left and right upper limbs,solving the problem in the asymmetric eight-segments impedance model,getting more accurate human impedance values and providing a new solution for quickly and accurately measuring body impedance values.Secondly,In order to solve the problem that the related parameters are numerous,highly non-linear,and serious related in the body composition measurement process,a selection algorithm of human body characteristic parameter based on Filter and clustering is proposed.this paper first uses the filter method to filter out the features that are not related to the body composition and then use the clustering method to remove the redundant feature to obtain a candidate feature set for human body component prediction modeling.At the same time,in order to measure the merits of the candidate feature set,the correlation analysis and regression analysis of candidate feature set and body composition were analyzed by SPSS.The results show that the proposed feature selection algorithm can select the optimal feature set related to the body composition,improve the fitting accuracy of the body composition prediction model,and reduce the prediction error of body composition.Finally,aiming at the problem that the model established by traditional statistical regression analysis method is poor adaptability,poor universality and low prediction accuracy,a prediction method of human body composition model based on support vector regression(SVR)is proposed.Firstly,SVR is used in the forecasting of human body composition to solve the problem that the sample size is limited and the prediction accuracy is low.Secondly,in order to give full play to the advantages of support vector machine in the small sample data,this paper presents the best sample set selection model.The model is composed of genetic algorithm(GA)and SVR.The model uses the global search ability of genetic algorithm to select the best small sample set from multiple sample sets,which is used to support vector machine regression modeling.The simulation results show that this method can effectively improve the prediction accuracy of human body composition model and solve the general problem of human body composition modeling in different areas.