The aging population has led to an increased prevalence of chronic kidney disease (CKD), associated with a higher incidence of gait disturbances and rise in fall rates. It is important that early detection and continuous monitoring of CKD to improve patient prognosis. Our study explores a non-clinical approach for detecting CKD by analyzing gait characteristics using inertial movement unit (IMU) sensors. With a deep learning approach, this research analyses gait measurement data from 276 individuals with varying stages of CKD and 217 healthy controls provided by Hallym University Chuncheon Sacred Heart Hospital. We propose a method for detecting CKD using a combined model of convolutional neural networks (CNN) and bidirectional long short-term memory (BiLSTM) networks, employing a normalized gait dataset. Our method achieved a binary classification accuracy of 84.98% in the segment approach and an accuracy of 79.61% in the voting approach. These results indicate the potential of using gait data for detecting the presence of CKD, signifying a new way towards early diagnosis and enhanced management of the disease.