Comparative Analysis of Deep Learning and Machine Learning Models for Early Prediction of Chronic Kidney Disease

Chronic Kidney Disease (CKD) is a progressive and frequently unidentifiable ailment presenting a considerable health issue in the world because of its sheer prevalence, comorbidity related to the illness, and financial effects. Early diagnosis of CKD is essential to help in early clinical intervention and alleviate the patient outcomes. Over the past years, there has been an increase in the artificial intelligence methods, especially machine learning and deep learning models, to aid in the early detection of diseases based on clinical data. The present review paper is a thorough performance discussion of machine learning and deep learning models that are applied to early CKD prediction. The paper embraces a positivist research philosophy and deductive research design since it focuses solely on secondary data collected in peer-reviewed literature. Thematic analysis is used as a method of empirical data synthesis as a systematic literature review methodology. The paper is a critical review of classic machine learning algorithms, such as Random Forest, Support Vector Machine, and Logistic Regression, and deep learning networks, such as Artificial Neural Networks and recurrent networks. The scientific findings provided in the literature are relatively compared with standard performance measures, such as accuracy, sensitivity, specificity, and area under the curve. The results show that deep learning models have high predictive power, however, the models based on the methods of machine learning ensembles tend to perform similarly with less computational complexity and higher interpretability. The author concludes that model selection must be a contextual issue by considering the available data, clinical applicability and the requirement of transparency.