Algoritmos e parâmetros imunológicos para predizer desfecho clínico da Covid-19

Abstract

The COVID-19 pandemic has impacted the world in several areas, such as health, economy and education. Currently, the circulating variant of SARS-CoV-2 is the Omicron (and its subvariants) which, despite its high transmissibility, is not as virulent as the Alpha, Gamma, Zeta and Delta variants. Each of these variants differs in terms of infection, symptoms and treatment. That's why we stress the importance of studying them. In this study, we analyzed clinical data using conventional statistics and machine learning models; ellipsometric measurements and analysis using machine learning models on serum samples from COVID-19 patients. We evaluated the different immune mediator profiles in patients infected with the different Alpha, Gamma, Zeta, Delta and Omicron variants using serum and tracheal aspirate. Our results showed that the use of conventional statistics was unable to identify possible early biomarkers of severity, while machine learning provided us with five possible biomarkers. Using the ellipsometry technique and machine learning, we were able to classify patients infected or not with SARS-CoV-2, as well as classifying the clinical outcome. These results demonstrate how different techniques can help in the diagnosis and prognosis of COVID-19 patients. About soluble mediators, we analyzed their production in the pulmonary and systemic compartments in the different variants. Our results showed that the production of soluble mediators differs between compartments and between variants, especially between the Alpha, Delta and Omicron variants. High levels of CXCL-10, CXCL-8, IL-6, TNF and INF- were found in the serum of COVID-19 patients, and in tracheal aspirates we found high levels of IL-13. Each of the variants has its own peculiarities that distinguish them from each other, and the way they interact with the immune system is also different. We evaluated the IgM, IgA and IgG Anti-N, Anti-S1, Anti-RBD and Anti-S2 antibodies and observed a different antibody production profile between serum and tracheal aspirate, also distinguished between variants. With this, we reinforce the importance of understanding the evolution of the virus and its interaction with the host and, for this, we can resort to artificial intelligence, techniques such as ellipsometry and the evaluation of soluble mediators and antibodies.