Diseño de un clasificador para pacientes en proceso de extubación

Abstract

The mechanical ventilation (MV) is a therapeutic strategy to mechanically assist or replace spontaneous breathing. With the objective of developing a software support for doctors was performed a study of respiratory signals using the discrete wavelet transform to determine the descriptors to indicate whether the patient can be disconnected from the mechanical ventilator. To reduce the dimensionality of the system was performed a principal component analysis (PCA), establishing three variables optimal, which are the inputs to the classifiers that were analyzed in the article: K-Nearest Neighbor and fuzzy logic.