Prediction of Chronic Kidney Disease with Artificial Neural Network

Good health and well-being is one of the goal of Sustainable Development. It specifically recognizes the growing concern posed by non-communicable diseases and strives to reduce premature deaths caused by these diseases by one-third by the year 2030. Chronic Kidney Disease (CKD) is a prominent non-communicable illness that substantially impacts the wellbeing of people worldwide, affecting around 10-15% of the global population and resulting in significant morbidity and mortality.In order to address the health complications related to CKD, which include hypertension, anemia, mineral bone disorder, poor nutrition, acid-base imbalances, and neurological problems, it is essential to intervene promptly with appropriate medications and accurately identify the stages of CKD. Several research studies have utilized machine learning techniques to detect CKD at an early stage, with a focus on early detection rather than precise stage prediction. Both binary and multiple classifications have been employed in these studies, utilizing the ANN algorithm as the predictive model. The study utilized feature selection methods including analysis of variance and recursive feature removal combined with cross-validation. The model’s performance was assessed through tenfold cross-validation, and the results of the experiments showcased the efficiency of ANN algorithm. With the best performance indicators, a total of four distinct machine learning-based classifiers—Decision Tree, K-nearest neighbour, Logistic regression, and Random Forest—have been studied. ANN achieved 0.983 accuracy, 0.98 precision, 1.0 recall, and 0.99 F1score.