Chronic Kidney Diseases Detection: A Machine Learning-Based Approach on Bangladeshi Patient Dataset

Chronic kidney disease (CKD) is a major concern for public health worldwide, impacting around 324 million people globally. It is particularly prevalent in South Asia, with a prevalence rate of 10%. Nevertheless, the deficiencies in healthcare infrastructure in less developed areas hinder the timely detection of early-stage CKD. As the condition worsens, the steady weakening of the immunological and neurological systems requires advanced treatments like dialysis. This study presents an important contribution by introducing a carefully curated dataset of CKD derived from patient reports. By utilizing a proper feature selection technique and SMOTE followed by a range of machine learning methods such as Logistic regression, KNN, SVM, and ensemble method, the proposed predictive model achieves prediction accuracy of 98.03 %, with high precision, recall, and Fl score. Significantly, this research is a new initiative in Bangladesh, since it is the first-ever endeavor to generate a dataset for CKD using Bangladeshi patient's data, thereby advancing the field of kidney disease research and diagnostics. Our study presents an important dataset and emphasizes the profound impact of data-driven approaches in healthcare. It provides an extensive resource for advancing research on CKD in Bangladesh and presents an innovative step towards global kidney disease diagnostics.