Background: The accurate evaluation of muscle mass by a non-invasive and easy method is the first step to help prevent falling events in elderly people. Methods: To develop greater predictive accuracy and precision in the measurement of body composition in lower limbs by bioelectrical impedance analysis (BIA), the Back Propagation Artificial Neural Network (BP-ANN) was used to calculate predictive results and was compared with data from dual-energy X-ray absorptiometry (DXA) in 22 male and 16 female elderly people in Taiwan. Fat-free mass (FFM), tissue weight, and fat mass (FM) of the lower limbs were directly measured by DXA, and the BIA values (Z) of left side hand to right side foot in the standing position were measured by BIA. The parameters of height, weight, age, gender and BIA values were combined to create the BP-ANN mathematical model, which was developed to predict the FFM and FM in lower limbs in elderly. Result: A relatively lower correlation coefficient (r 2) of 0.964 and standard deviation (2SD) of 0.01 ± 3.64% were obtained for the prediction of FFM and FM by BIA with the BP-ANN mathematical model, whereas the linear regression analyzing model had an r 2 value of 0.845 and 2SD of 0.12 ± 7.68%, respectively. The performance of the BP-ANN mathematical model at BIA measurement was superior to that of the current linear regression model. Conclusion: In summary, the greater predictive accuracy and precision made the application of BIA with the BP-ANN mathematical model more feasible for the clinical measurement of FM and FFM in the lower limbs of elderly people. Copyright © 2012, Taiwan Society of Geriatric Emergency & Critical Care Medicine. Published by Elsevier Taiwan LLC. All rights reserved.
Liu, T. P., Kao, M. F., Jang, T. R., Wang, C. W., Chuang, C. L., Chen, J., … Hsieh, K. C. (2012). New application of bioelectrical impedance analysis by the back propagation artificial neural network mathematically predictive model of tissue composition in the lower limbs of elderly people. International Journal of Gerontology, 6(1), 20–26. https://doi.org/10.1016/j.ijge.2011.09.025