Research on the Impact of Gradient Boosting Decision Trees and Dynamic Weights on Maternal Physical and Mental Health on Infants

In this paper, the concept of ensemble learning is first utilized to combine multiple classification and regression models. The mother's gestational age, EPDS, and maternal age are obtained through supervised learning and then examined for their correlation with infant outcomes. Subsequently, the infant's behavioral characteristics are trained, and the accuracy of this model is 64.8. Furthermore, a dynamic programming model is employed to determine the treatment costs required for transforming infant behavioral characteristics into intermediate and calm types, which amount to 40271 yuan and 44277 yuan. Lastly, utilizing dynamic weighting, a comprehensive evaluation of the infant's overall sleep quality is conducted, followed by prediction of the infant's comprehensive sleep quality using a random forest model, achieving a high prediction accuracy of 76.2%. Through the dynamic programming model, it is determined that to elevate the infant's comprehensive sleep quality to excellent, a treatment cost of 42532 yuan is required.