Comparativa de modelos multivariados basados en aprendizaje automático para la predicción del riesgo de diabetes en etapas tempranas
Keyword(s):
Data Set
◽
Diabetes is one of the fastest-growing, life-threatening, chronic degenerative diseases. According to the World Health Organization (WHO), it has affected 422 million people worldwide in 2018. Approximately 50% of all people who suffer diabetes are not diagnosed due to the asymptomatic phase which usually lasts a long time. In this work, a data set of 520 instances has been used. The data set has been analyzed with the next three algorithms: logistic regression algorithm, decision trees and random forest. The results show that the decision tree algorithm had better performance with an AUC of 98%. Also, it was found the most common symptoms that a person with a risk of diabetes presents are polyuria, polydipsia and sudden weight loss.