Testing And Research Of Body Composition-based On Bioelectrical Impedance

Human body composition is the ratio among various compositions interior the body. It is of important value in clinical medicine and basic research. It can monitor the body's nutritional and fluid status, provide the normal range of the body composition and evaluate the growth and development of the body and so on. So, composition test is an important item of health check.There are many ways to test human body composition. The way based on bioelectrical impedance is an indirect means to evaluate body compostion. It's principle is as follows. First, we import an AC current into the human body. Then, we will obtain bioelectrical impedance of various components by measuring the current. By analyzing, we will get the composition of corresponding part. This method is widely used because it has the features of easy and non-invasive.In this thesis, we analyse the development of human body composition test system. Based on the principle of bioelectrical impedance, we design a body composition testing and evaluation system. The main work we complement is as follows.1. We have finished the design of the system according to the multi-segment human body model. The system uses eight electrodes, single-frequency measurement techniques. It is a device with features of high precision, good reproducibility. We can get bioelectrical impedance of various components through analsing software. Then, we could composition of various compositions by analyzing model. In the end, we can finish the health evaluate.2. We have improved the modeling method to deal with body composition data. With the purpose of how to improve the prediction accuracy of human body composition, that is, to improve prediction accuracy and generalization ability, we study two modeling methods in this thesis. They are the method based on the statistical regression and the one based on support vector machine.We can get a series of prediction equation in statistical regression. Since equations can only sustain the commen features of training sample, discarding the other personality characteristics, the generalization ability is limited. Analyzing statistical theory, we introduce statistical theory. Then, we propose replace traditional prediction equations with prediction model files. Sine this file sustains almost all features of the training sample, it can adapt to different test groups. Meanwhile, this method can get the global optimal solution making use of convex quadratic programming method and avoid the problem of local optimal solution. It is a major feature of this paper that we make use of support vector machine to establish the body composition model.We obtain the law that body composition varies with age and living habits by analyzing sample data. We do a primary study at how to establish human body health expert system with the body composition data. We mainly focus on how to develop an appropriate exercise prescription based on human body composition.At present, the body composition test can only pay attention to test result at some time. But the test result at a certain time is only a static value which is not enough to reflect the trend of human health. We propose, in this paper, that we should establish a human health data management system. This system could help us to manage healthy data dynamicly. According to the human body health trend reflected by history data, it could give a reasonable exercise prescription and ntrition formula. Finally, these programs could make the ratios among the body compositions optimal by adjusting continually.