Machine Learning-Based Monitoring and Prognosis of Chronic Kidney Disease Patients

The study is aimed at examining the use of ML techniques for detecting and forecasting CKD progression. Experimental models were constructed utilizing the whole datasets including demographic information, medical history, laboratory findings, and clinical notes; then, they were assessed. The ML algorithms embedded are decision trees, random forests, support vector machines, and gradient boosting. The experimental results showed clear evidence of good predictive accuracy across all algorithms, with gradient boosting achieving the highest accuracy of 90 percent. Besides that, the precision, recall, F1-score, and the area under the receiver operating characteristic curve (AUC-ROC) were assessed, having values from 0.82 to 0.92. The performance of the proposed ensemble method which optimizes both two time-varying Markov chains among the related works is superior to other methods. The studied variables, such as serum creatinine, glomerular filtration rate, age, blood pressure and urine protein levels were linked to the progression of CKD in the variable importance analysis. This emphasizes the ability of ML-based methods in providing for CKD patients’ care by the means of early diagnosis, personalized medicine, and improved clinical results (outcomes).