Redes neuronales y regresión logística técnicas predictivas de la mortalidad en medicina interna del HRDT

• Autor: Guevara Aguilar, Victor Hugo
• Formato: tesis de grado
• Fecha de Publicación: 2017
• Institución: Universidad Nacional de Trujillo
• Repositorio: UNITRU-Tesis
• Lenguaje: español
• OAI Identifier: oai:dspace.unitru.edu.pe:20.500.14414/10638
• Enlace del recurso: https://hdl.handle.net/20.500.14414/10638
• Nivel de acceso: acceso abierto
• Materia: Redes neuronales, Regresión logística

Statistics today have become one of the most needed sciences and used in different_x000D_ fields of work because of its way of applying its various techniques for obtaining results_x000D_ and appropriate decision making. Within these techniques we have the binary logistic_x000D_ regression, a technique that allows to verify the causal relations of a variable when it is_x000D_ nominal, these techniques are applied to the health sciences allow us to analyze the_x000D_ results in explanatory and predictive terms for To evaluate mortality in hospitals, this_x000D_ technique and others are apparently not as efficient as the technique of neural networks,_x000D_ since this technique is not necessary to evaluate the assumptions of normality, since_x000D_ they are considered as statistics, nonparametric tests. That is why in previous studies it_x000D_ is stated that they are much better than the regression techniques, it is for this reason_x000D_ that the present study aims to determine which of the two techniques is better to predict_x000D_ hospital mortality in internal medicine of HRDT. Taking into account the variables that_x000D_ can influence this mortality, which are the sex, age and region of origin. In order to_x000D_ fulfill our objective, we obtained the database of the patients of the area of internal_x000D_ medicine of the HRDT; For the statistical analysis, the logistic regression technique and_x000D_ neural networks were used, And for these, the technique that best predicts is the_x000D_ technique of logistic regression, since, at the moment of observing the results in terms_x000D_ of the classification percentages, the difference is notorious