Improving Accuracy in Chronic Kidney Disease Diagnosis: A Machine Learning Model for Overlooked Patients

Timely diagnosis of chronic kidney disease (CKD) is critical for improving patient outcomes and reducing healthcare costs. However, conventional diagnostic methods are often complex and costly, limiting reliance on glomerular filtration rate measurements alone. This study aims to develop a machine learning model that utilizes routine test data to aid in diagnosing CKD in patients who have missed standard testing. A machine learning model was developed and evaluated using routine test data. The model's performance was assessed through the area under the receiver operating characteristic curve, diagnostic accuracy, precision, and sensitivity. An external validation set was utilized to evaluate the model's application in clinical practice over a specific time span. The machine learning model achieved an area under the receiver operating characteristic curve of 0.90. In the external validation set, the model correctly diagnosed 85.21% of patients with missed chronic kidney disease. This study is the first to demonstrate the use of routine test indicators to assist in diagnosing chronic kidney disease in patients who have missed testing. The findings highlight the potential added value of this approach in clinical practice, offering a feasible solution to improve early diagnosis of chronic kidney disease.