A Bioelectrical Impedance Muscle Mass Prediction Model was Established Based on AOA-BP

In this paper, a muscle mass prediction method based on bioelectrical impedance (BIA) was proposed to address the problems of high cost, complex measurement method, low accuracy, and inapplicability of muscle mass measurement methods. In this method, eight electrodes were used to detect the muscle mass of the left and right upper limbs, lower limbs and trunk of the human body. The voltage signals of the above five parts were collected by high frequency current (120kHz) and low frequency current (20kHz). The laboratory-defined hardware equipment is used to process the signals and obtain the required impedance values. At the same time, 30 dimensional characteristics of other human parameters such as waist-to-hip ratio and heart index were collected. A prediction model for the best subset of features of samples was established and verified by using the Improved Marine Predator Optimization Algorithm(MPA-DMA) to reduce the dimension to 18 and the Arithmetic Optimization Algorithm (AOA-BP) to optimize the BP neural network, thereby completing the prediction of individual muscle mass. Experimental results showed that the proposed method, which combines more relevant human parameters, could significantly enhance the prediction accuracy of the muscle mass prediction model to 98.9%.