In recent years, Nephrotic Syndrome (NS) has emerged as a significant challenge in global public health. NS encompasses a spectrum of diseases affecting kidney function, with Membranous Glomerulonephritis (MGN) being the most prevalent in mammals.Common indicators of NS include protein loss in urine and decreased blood albumin levels. Primary glomerulonephritis is the leading cause, with Anti-Phospholipase A2 receptor antibodies (anti-PLA2R) serving as a diagnostic marker for MGN, as per Kidney Disease: Improving Global Outcomes (KDIGO) guidelines. Our study employed various machine learning techniques to develop predictive models using disease classification data and examination results such as complete blood count, blood and urine biochemistry. After data refinement, the dataset was split into training and testing sets, with Python utilized to build machine learning models. Statistical analysis revealed significant differences (p<0.05) in BUN, Creatinine, and eGFR between normal individuals and those predicted to have the disease. The machine learning process incorporated Synthetic Minority Oversampling Technique (SMOTE) to address imbalanced data and 5-Fold Cross-Validation for validation. Final models utilized the Random Forest algorithm and eXtreme Gradient Boosting (XG Boost), achieving an Accuracy and AUC of 0.83, indicating robust predictive performance. Our methodology, leveraging routine blood and urine tests alongside machine learning algorithms, enables accurate prediction of MGN. In the future, this methodology holds promise for application in mouse research and potentially diagnosing human diseases in clinical.