TITLE: Detection of asynchronies in mechanical ventilation by predictive analysis:
Applications in Machine Learning
author(s): Assis, C.C., Ferreira, B.S., Silva, G.C., Sousa, E.S., Sousa, M.R. & DelMonaco, A.D.M.
ABSTRACT: This study aims to present the detection of asynchrony in mechanical ventilation through predictive analysis using machine learning technology. In which the mechanical ventilator is responsible for the improvement of gas exchange and the reduction of respiratory difficulty presented by the patient. With the use of a mechanical ventilator, a relatively common phenomenon can occur, known as patient ventilator asynchrony (PVA), where there is a mismatch between the ventilatory needs demanded by the patient's respiratory center. For this, a literature review on the topic was carried out, which refers to a case study through materials and articles accessed in the virtual environment with the help of research platforms such as Google Academic, Scielo and Pubmed, so that the asynchrony could be analyzed. ventilator patient (VPA) in mechanical ventilator equipment (MV), its classification and management, as well as the interaction for the knowledge of health professionals for a better identification of asynchrony, with the incremental possibility of using machine learning technology (Learning of machine). Therefore, the graphical introduction by machine learning is one of the premises of this article, so that predictive analyzes help the nurse or clinical staff to detect and distinguish more quickly and practically the occurrences of asynchrony, identifying the patterns and having the solutions so that it can better assist the patient in a moment of crisis. As a result of the article, the implementation of the reported analysis models can help to fight the new coronavirus pandemic (COVID-19), since it is possible to predict the necessary configurations for these patients, once with the projected model and in effective functioning, a challenge to be described is the development of the machine learning method applicable in the software of current fans and also a model with the necessary processing capacity for this technology.
KEYWORDS: Big Data, double trigger, deep learning, late cycling, ineffective effort.