Regresión logística para la identificación de factores pronósticos de mortalidad en pacientes con cáncer de mama en SOLCA guayaquil, periodo 2015-2020

Abstract

The main objective of this research work was to identify those prognostic factors that influence the mortality of patients diagnosed with breast cancer, based on pre-established factors, analyzing their significance and direct influence; as well as relationships that allow generating predictive models that anticipate the behavior of the Disease in a patient. The development of the work was carried out at the National Oncological Institute “Dr. Juan Tanca Marengo” (SOLCA Guayaquil), based on the data that rest in the institution´s own system. All patients diagnosed with breast cancer in the period 2015-2020 were taken into account, with a total of 3062 patients, working only with 956 of them after correct data filtering. A mixed investigation was carried out, combining qualitative and quantitative variables, studied through predictive models modeled from the same logistic regression that was carried out in the statistical software R. In addition, it is of a theoretical type, encompassing a descriptive and explanatory investigation, of a non-experimental nature, based on a hypothetical-deductive inference. With 95% confidence, three models were proposed, which were significant; with an acceptable accurate ability to predict (greater than 75%), defining the three models as prognostic factors influencing size, weight, time of evolution, tumor size, stage, and the states that form the stating (T, M, N). In addition, with approximately acceptable probabilities, model one is defined as the most suitable for predicting the behavior of a patient before the disease. The study of prognostic factors is an important topic to deal with in the future with a bigger number of patients, in order to achieve more precise models with a greater predictive capacity, which allow us to define more influential and decisive prognostic factors.