Redes neurais aplicadas na parametrização da ventilação mecânica de pacientes com insuficiência respiratória aguda

Abstract

Mechanical ventilation substitutes totally or partially spontaneous ventilation and is indicated in the accute breathing insufficiency or accute chronic insufficiency. The mechanical ventilation provides betterment in the gas exchanges and diminishment of breathing work and can be used in a noninvasive form through an external interface, normally a mask, and in an invasive form through and endotracheal tube or a tracheostomy cannula. Given these facts, for the betterment of the patient, frequent manual adjustments must be done to the mechanical ventilator, according to determined reactions or signals observed and obtained from patients by the specialist in intensive care, implying in a constant and exhausting work. This dissertation has the objective of presenting a solution capable of automating the setting of parameters of the mechanical ventilator, regarding software, by the application of Artificial Neural Networks (ANN), and consequently improve the treatment of the patient, with the intention of enhancing human resources of health. On the tests, the methodology employed was efficient on the automatic setting of parameters of the ventilator, obtaining a success rate of 100% in the quantitative evaluation and of 79% in the qualitative evaluation, carried out by a group of physical therapy specialists. The study showed the viability of the employment of the method in helping the treatment of the patients, as well as in the optimization of human resources of health