Chronic kidney disease (CKD) is symptomless in the early stages, and it is difficult to diagnose. In order to increase the precision of CKD diagnosis and staging, a hybrid model that integrates a 1D Convolutional Neural Network (CNN) and Random Forest Algorithm (RFA) is proposed in this article. Our model demonstrated excellent performance with 95% accuracy in binary classification in the existence of CKD for a patient and 92% accuracy in multi-stage classification. RFA is a challenging technique for the early identification of CKD because of its capability in multiclass classification and CNN’s capacity to identify complex patterns. The results prove the potential of the hybrid models in clinical settings and deliver healthcare professionals with a consistent way to quickly intervene and develop customized treatment regimes.