Predicción de la evolución hacia la hipertensión arterial en la adultez desde la adolescencia utilizando técnicas de aprendizaje automatizado
Fecha
2013-07-04
Autores
Alfonso González, Wilfredo
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
Las recomendaciones de la Sociedad Europea de Hipertensión Arterial reconoce que la hipertensión
arterial en la edad pediátrica es un problema médico que ha ido incrementándose con repercusiones
negativas presentes y futuras; y hace referencia a la ausencia de estrategias o políticas de salud
organizadas que enfrenten eficazmente la enfermedad en este ámbito. Existen estudios sobre el
pronóstico de la conversión de un adolescente prehipertenso en un adulto hipertenso en Cuba. Sin
embargo, los índices de exactitud obtenidos son aun bajos, ya que solo se alcanza el 80% de casos
bien clasificados; de ahí que es necesario seguir trabajando la predicción, con mejor exactitud, de
adultos hipertensos a partir de adolescentes prehipertensos. El objetivo de este trabajo consiste en
aplicar técnicas de aprendizaje automatizado que permitan pronosticar la conversión de un
adolescente prehipertenso en un adulto hipertenso en Cuba, elevando los índices de exactitud. Los
principales resultados son: (1) Se seleccionaron las técnicas de selección de rasgos, edición de
conjuntos de entrenamiento, y de aprendizaje automatizado, a aplicar que puedan contribuir a
obtener mejor exactitud en la clasificación de adultos hipertensos a partir de adolescentes
prehipertensos, (2) se aplicaron los algoritmos para el preprocesamiento de los datos y de
aprendizaje automatizado para pronosticar la hipertensión en la adultez a partir de adolescentes
prehipertensos, y (3) se realizó una comparación estadística de los clasificadores aplicados al conjunto
de pacientes prehipertensos estudiados, obteniéndose un 88% de clasificación correcta al aplicar
técnicas de edición para problemas con clases desbalanceadas al conjunto de datos.
The recommendations of the European Society of Hypertension recognizes that hypertension in children is a medical problem that has been increasing with negative impacts present and future, and refers to the absence of health strategies or policies that address effectively organized disease in this area. Studies on the prognosis of the conversion of a teenager in an adult hypertensive prehypertension in Cuba. However, the obtained accuracy rates are even lower, as it only reaches 80% of cases correctly classified, hence the need to continue working the prediction, with better accuracy, hypertensive adults from prehypertensive adolescents. The objetive of this work is to apply machine learning techniques that will predict the conversion of a teenager in an adult hypertensive prehypertension in Cuba, raising accuracy rates. The main results are: (1) were selected feature selection techniques, editing training sets, and machine learning, to apply that can contribute to obtain better accuracy in the classification of hypertensive adults from adolescents prehypertension (2) algorithms were applied to the data preprocessing and machine learning to predict hypertension in adulthood from prehypertensive adolescents, and (3) there was a statistical comparison of classifiers applied to all prehypertensive patients studied, obtaining 88% correct classification by applying editing techniques for problems with unbalanced classes to the data set.
The recommendations of the European Society of Hypertension recognizes that hypertension in children is a medical problem that has been increasing with negative impacts present and future, and refers to the absence of health strategies or policies that address effectively organized disease in this area. Studies on the prognosis of the conversion of a teenager in an adult hypertensive prehypertension in Cuba. However, the obtained accuracy rates are even lower, as it only reaches 80% of cases correctly classified, hence the need to continue working the prediction, with better accuracy, hypertensive adults from prehypertensive adolescents. The objetive of this work is to apply machine learning techniques that will predict the conversion of a teenager in an adult hypertensive prehypertension in Cuba, raising accuracy rates. The main results are: (1) were selected feature selection techniques, editing training sets, and machine learning, to apply that can contribute to obtain better accuracy in the classification of hypertensive adults from adolescents prehypertension (2) algorithms were applied to the data preprocessing and machine learning to predict hypertension in adulthood from prehypertensive adolescents, and (3) there was a statistical comparison of classifiers applied to all prehypertensive patients studied, obtaining 88% correct classification by applying editing techniques for problems with unbalanced classes to the data set.
Descripción
Palabras clave
Hipertensión Arterial, Pronóstico de Enfermedades, Técnicas de Aprendizaje Automatizado, Clasificación