Linear Regression Models for Total Body Water Measurement Based on Bioelectrical Impedance Analysis

This work proposes new linear regression models for total body water (TBW) measurements for both male and female human being based on bioelectrical impedance analysis. In this research total 2817 (1397 male and 1420 female) data have been used. Age, height, bioelectrical impedance at 100 kHz frequency, body mass index (BMI) have been used for the development of mathematical models. The proposed models have been analyzed statistically and the results show that the correlation (Pearson) coefficients are 0.998 (p<0.001) for male and 0.997 (p<0.001) for female individuals which denote excellent matching with actual data. Besides intervals of LOA is only -1.36 L to 1.59 L and -1.17 L to 1.12 L for male and female data respectively and most of the data follows such a small limit of agreement. The absolute errors (mean ± Standard Deviation) are (0.64 ± 0.41) L and (0.48 ± 0.32) L for male and female data respectively whereas the bias are 0.11 L for male and -0.02 L for female population only. The RMSE are also very low and which are 0.76 L for male and 0.58 L for female people. Comparing the results of this research with existing models it is seen that the proposed models can be more suitable for TBW measurement.