Study On Human Body Composition Analysis Method Based On Bioelectrical Impedance Technique

Body composition analysis understands the basic health status of the human body by analyzing the components and their proportions that make up the human body.It is increasingly widely used in the fields of disease prevention and diagnosis,rehabilitation evaluation,and fitness guidance.The bioelectrical impedance technology has the characteristics of non-invasive,harmless,low cost and simple operation,and it is the basic measurement technology widely used in human body composition analysis instruments.At present,most human body composition analysis instruments based on bioelectrical impedance technology use empirical formula models,and the predictcion results are not accurate.This thesis focuses on the study of body composition analysis methods based on bioelectrical impedance measurement combined with intelligent algorithms.The main research contents are as follows:Introduced the method and theory of bioelectrical impedance measurement technology,as well as the current development of the body composition prediction model based on the bioelectrical impedance method,and analyzed the limitations of the empirical formula model established by the multiple linear regression method used in the current body composition analysis.With the Red-P itaya FPGA development board as the core,an eight-electrode bioelectrical impedance spectroscopy(BIS)measurement system for body composition analysis is designed,and the signal excitation module,response acquisition module and analog switch module are introduced in detail.The system uses multi-frequency sine(multisine)signals as the measurement excitation,which can realize the simultaneous measurement of human body BIS at 256 frequency points in the range of 1.91kHz to 950.81kHz;repeat 50 measurement experiments with standard three elements,and the amplitude standard deviation is below 1.41Ω The phase standard deviation is below 1.98°;through error comparison with the SFB-7 instrument,the absolute error of the im pedance amplitude is less than 1.45Ω,and the absolute error of the phase is less than 1.80°;the stability and accuracy of the BIS measurement system are good.Principal component analysis method is used to reduce the dimensionality of human BIS data.The data dimension is reduced from 256*6 to 256*1.After dimensionality reduction,the data contains more than 99.82%of the original impedance amplitude and more than 99.28%of the phase;BP neural network is used respectively,GA optimized BP neural network and support vector regression machine three intelligent algorithms to establish a body composition prediction model;through the comparison and analysis of the prediction results of the three models and the measurement results of the SFB-7 instrument,the GA optimized BP neural network prediction results are better than the other two models;Compared with the measurement results of the SFB-7 instrument,the prediction results show that the relative error of intracellular fluid(ICW)is less than 4.97%,the relative error of extracellular fluid(ECW)is less than 4.69%,and the relative error of total body water(TBW)is less than 3.82%,the relative error of non-fat mass(FFM)is less than 3.60%.