The occurrence of type 2 diabetes mellitus, a chronic metabolic condition, has been continuously increasing worldwide. The World Health Organization and Diabetes Association (DA) standards consider both clinical and laboratory factors and provide the framework for screening and diagnosis. The nerve disorder has no known cure yet. Pre-diabetes and morphological diabetes patients’ impacts are yet apparent. The prolonged undiagnosed diabetes mellitus may cause major nerve disorders. It leads to severe nerve impairment that affects the brain and spinal cord which led peripheral neuropathy. This research model is designed to evaluate the clinical applications of percussion entropy index (PEI) and electrophysiological findings (EPF) for predicting the severity of nerve damage in patients. The dataset of five hundred and sixteen predefine diabetes patients were classified into mild, moderate, and severe based on clinical index values. Different machine learning prediction models like logistic regression (LR), support vector machine (SVM), light gradient boosting (LGB), and random forest (RF) are used to perform comparative analysis with the proposed hybrid model. The output of this research study shows that the Hybrid RF model was very effectively used to predict the severity level of diabetic neuropathy.