Artificial Intelligence Identification Model for Chronic Kidney Disease

Many people suffer from Chronic Kidney Disease (CKD). Nowadays, CKD is one of the top ten causes of death. CKD should via invasive examination to understand participants health status. If a non-invasive identification model can be established, it can provide users with self-assessment which let users quickly understand their physical condition. This study used machine learning method to establish an identification model of Chronic Kidney Disease. This study found the associated factors with kidney failure from the literature. Selected MJ database as information resources. Used two different factor selection methods to training model. Compared the performance with K-Nearest Neighbor, Support Vector Machine, Logistic Regression, Artificial Neural Network, Decision Tree, Random Forest, eXtreme Gradient Boosting and Vote Algorithms, used the better one to establish the model. In this study, the best model used Vote algorithm to establish the model, and can only use 13 non-invasive factors. The accuracy is 88%, the precision is 73%, the sensitivity is 69%, the specificity is 93%, and the AUC is 0.92. The contribution of this study is to use non-invasive factors to identify Chronic Kidney Disease, but it is a preliminary evaluation and ultimately requires doctors to diagnose. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.