Early Stage Kidney Chronic Diseases Diagnosis using Feature Fusion and Deep Learning Algorithms

Chronic Kidney Disease (CKD) is a global health issue due to pre-existing conditions like anemia, diabetes, and hypertension. Early warning signals are commonly neglected, complicating diagnosis and treatment. This study uses high-performance data mining techniques to provide self-analysis and prognostic tools for medical professionals. Clinical decision support using domain experts and specialists aids CKD prognosis and rehabilitation. We use advanced data mining to evaluate demographic characteristics, test findings, and patient medical information to uncover insights and patterns. We used a modified artificial neural network to extract characteristics from a dataset and long-term and short-term memory to classify chronic kidney disease stages. LSTM algorithms improve the accuracy of CKD diagnosis and stage prediction. Initial validation tests demonstrate a 92.23% accuracy gain over traditional methods. The binary classification AUC was 99.95% and the multi-class classification was 94.13% with the proposed technique. LSTM algorithms use temporal dynamics in patient data to enhance diagnostic accuracy, early diagnosis, therapy customization, and healthcare costs. We proposed a novel CKD diagnostic and prognosis paradigm and test multiple classification algorithms for disease diagnosis and staging. Multiple methods improve reliability and account for CKD diagnostic peculiarities.