Regression models in R for predicting birth weight and type of delivery

Abstract

The purpose of this research was to perform a predictive statistical analysis of live births in Guadalajara de Buga, Colombia, using R. The aim was to achieve an accurate prediction of birth weight based on the length of the fetus, as well as the type of delivery based on the weeks of gestation. The methodology included the construction of several regression models, starting with a simple linear regression between 'Weight' and 'Height'. Subsequently, a multiline regression was implemented to incorporate 'Gestation Time' as an additional predictor. Finally, a logistic regression was carried out to predict the type of delivery according to the weeks of gestation. The results revealed a positive linear correlation between 'Height' and 'Weight', indicating a viable linear regression model. The resulting equation made it possible to predict the average weight gain for each additional centimeter in 'Height'. The multiline regression showed a coefficient of determination of 62.8%, with 'Height' and 'Gestation Time' as significant variables. Logistic regression provided valuable information on how 'Gestation Time' affects the probability of having a cesarean delivery. The research established precise relationships between selected live birth variables, providing valuable predictive tools through regression and logistic models. These results have the potential to positively impact clinical decision-making and obstetric planning, significantly contributing to the improvement of maternal-neonatal care.