Assessment and prediction of Chronic Kidney using an improved neutrosophic artificial intelligence model
Neyda Hernández Bandera1, Jenny M. Moya Arizaga2, Enrique Rodríguez Reyes2
1Docente de la carrera de Medicina de la Universidad Regional Autónoma de los Andes (UNIANDES), Ecuador
2Docente de la carrera de Medicina de la Universidad Regional Autónoma de los Andes (UNIANDES Santo Domingo), Ecuador
3Docente de la carrera de Medicina de la Universidad Regional Autónoma de los Andes (UNIANDES Ambato), Ecuador
Emails: ua.neydahernandez@uniandes.edu.ec; docentetp54@uniandes.edu.ec; ua.enriquerodriguez@uniandes.edu.ec
Abstract
CKD, or chronic kidney failure, is characterized by a gradual decline in kidney operation over time and may be linked to a wide range of medical conditions. Initial detection and therapy are the best tools for combating chronic kidney disease, although they often only delay the development of renal failure. The eGFR-based CKD grading system is useful for risk stratification, patient monitoring, and treatment strategy development. Personalized care and treatment planning will be possible if this research is successful in predicting how soon a CKD individual will need to begin dialysis. The machine learning methods used to predict CKD. But the dataset contains uncertain information, so the neutrosophic set is used to overcome this issue. This paper suggests a framework including the neutrosophic DEMATEL and machine learning method to predict CKD. The neutrosophic DEMATEL method is used to give weights to all variables of the dataset. Then conduct the preprocessing data to eliminate the variables with the least weight. The three machine learning methods used in this paper are Gradient Boosting (GB), Ada Boosting (AB), and Random Forest (RF). The results show the accuracy of the three algorithms. The AB has a 99.166% accuracy, and it is the highest accuracy in this paper followed by the GB has 98.3%, then RF has 92.85%.
Keywords: Neutrosophic Set; DEMATEL; Machine Learning; Kidney; Prediction