Evaluación de modelos matemáticos para estimar el peso y talla en pacientes adultos usando CRM, RMSE, Pearson y Bland Altman

Introduction: The availability of anthropometric data (weight and height) of patients with little or no mobility are important for medical and nutritional treatment, to estimate these values mathematical models that reproduce with greater fidelity have been used, so it is important to evaluate the model estimation method. Objective: To Assess the mathematical models of Rabito, Chumlea and HNHU to estimate weight and height in adult patients using the ERM, RMSE, Pearson and Bland Altman methods. Materials and methods: Data from 31 patients between 20 and 65 years old are considered. The data were knee height (RA), arm circumference (AB), abdominal circumference (AC), calf circumference (CP), mean arm length (MB), and arm span (EB) comprised of eight Rabito models for estimate weight and height, four from Hospital Nacional Hipólito Unanue (HNHU) and four from Chumlea. The quality of the estimation was evaluated by the Pearson Correlation, Relative Mean Error (ERM), Root Mean Square Error (RMSE) and Bland Altman methods. The level of association between the methods was determined by Pearson. Calculations were developed using R 4.1.0 statistical software. Results: The measurements by the Pearson method present a variation of 54%, the ERM method of 26.65%, by Bland Altman of 8.49% and RMSE 6.1%. The RMSE and Bland Altman methods present an association of 0.72. The Rabito 3M (RMSE=4.38) and Rabito 3F (RMSE=4.36) models reproduce the weight values with greater fidelity and for height estimation the Rabito 2M (RMSE=3.64) and Rabito 2F (RMSE = 3.82) models. Conclusions: The RMSE and Bland Altman methods have a good association, presenting good stability in the evaluations. Rabito's mathematical models have good estimates for weight and height.