Diseño de una aplicación web con integración a Big data y Machine Learning para el análisis predictivo de emergencias en pacientes con enfermedades crónicas de la unidad Hospitalaria clínica San Francisco

Abstract

Abstract: This research work demonstrates the advantages of using technologies in data processing, in order to obtain results for decision making, for this case tools such as Hadoop, Rstudio and Weka were applied to process and predict related information. to chronic diseases, giving more emphasis to diabetes being a chronic disease on the rise according to the OMS. The development of this degree work includes the chapters referring to the theoretical support of the technologies to be used, the research methodology, the application of Hadoop as Big Data that gave positive results since it allowed observing the advantages of being able to apply a pseudo-distributive model saving costs, without the need to invest in a lot of equipment. The use of Machine Learning with Weka, made it possible to observe that the decision tree model such as the Bayesian applied to the data of diabetic and non-diabetic people gave the same results, obtaining a return of predictive accuracy of 86.88%, this implies that, of 10 processed data, approximately 8 are correct. Finally, a web tool was developed following a basic SCRUM methodology that integrates the results obtained with Big Data and Machine Learning, providing positive results for the management and control of hospital information, such as access to patients to report incidents or emergencies that allow them to be attended to quickly.