PREDICTING THE IDEAL WEIGHT IN THE PROCESS OF DIALYSIS IN CHILDREN USING REGRESSION ALGORITHMS

Abstract

This study shows the use of regression algorithms in predicting the ideal weight of children during the process of haemodialysis. The data regarding height and weight for calculating the BMI (Body Mass Index) are collected at the University Children’s Hospital Tirsova in Belgrade. In addition, bioimpedance, haematocrit and blood pressure are measured. The collected data are pre-processed and transformed before the application of regression algorithms. The application of regression algorithms predicts the ideal weight of the patient, which, in practice, is determined by the attending physician. Two types of errors are used for the evaluation: mean absolute error (MAE) and root mean squared error (RMSE). The results indicate satisfactory accuracy, while future work refers to the implementation of algorithms through a smart watch which would enable the timely receipt of the information regarding the ideal weight.