Aprendizaje Computacional Para La Estimación Automática Del Peso Al Nacimiento Usando Variables Multimodales Materno-Fetales

Resumen

During pregnancy there are various factors that directly or indirectly affect maternal and child health, computational learning is presented as a pre-diagnostic tool in conjunction with the observation of obstetric experts in the early prediction of birth weight through characteristics of biometric origin. Through the 14 characteristics obtained by 584 voluntary participants, the use of a non-parametric regression algorithm is established, based on Gaussian processes. With an 8.3% weighted mean absolute error, the early prediction of birth weight is presented, allowing medical personnel and obstetric experts to handle information for the care of the mother-child population.