Prediction model of human body composition based on physiological information entropy

Abstract

Body composition analysis can not only reflect the state of health, but also can play a role in disease prevention. Aiming at many influencing factors, complex modeling issues of the existing bioelectrical impedance analysis algorithms, this paper draws information entropy theory into modeling the human body composition for the first time, establishes entropy evaluation criteria of physiological characteristic parameters, puts forward feature selection algorithm based on physiological information entropy, selects a reasonable subset of features that can most effectively interpret body physiological information and have a minimal number of features to give the body composition prediction fitted model. Experimental results show that the algorithm can select the useful characteristic parameters and the fitted model improves the accuracy of body composition prediction.