Objective. Construction and verification of a multidimensional mathematical model for predicting the obesity in school-age children. Material and methods. The 84 obese children aged 8 to 15 years were examined. The control group consisted of 36 healthy children. Of the 53 analyzed risk factors for the formation of obesity in children, the most informative ones were selected; based on that the decision rules were built in the form of classification trees. Results. The most significant indicators for the prognosis of obesity were the following indicators: physical activity and the frequency of snacks between meals in the child, the mother’s body mass index and the duration of breastfeeding. All these factors show an influence on the obesity formation, even in a one-dimensional case with a sensitivity of at least 75% and a specificity of at least 60%. With a multivariate assessment, the forecast results are much more accurate: the sensitivity and specificity are not less than 80%. The correctness of the obtained forecast models was checked by the cross-validation method while the probability of classification error R did not exceed 20%. Assessment of predisposing risk factors already in the first year of a child’s life makes it possible to give a reliable prognosis about the possibility of developing obesity at an older age. Conclusion. The use of the proposed prognosis model in clinical practice will contribute to the adequate formation of high-risk groups, the choice of individual monitoring and prevention programs for this group of patients by pediatricians and pediatric endocrinologists.
Level of Research Output
- VAK List