A Comparative Analysis of Classical Machine Learning and Deep Learning Approaches for Diabetic Peripheral Neuropathy Prediction

Abstract

The most prevalent chronic complication of diabetes mellitus, diabetic peripheral neuropathy (DPN), has become an increasingly prominent public health issue. Diabetic associated complications, which affects all major organs of the body, are common Diabetes Mellitus (Liu et al. in PLoS ONE 14:1–16, 2019 [1]; Jelinek et al. in J. Diabet. Complicat. Med. 01:1–7, 2016 [2]). In this analysis, the traditional machine learning algorithms (MLA) and deep learning method Multi-Layer Perceptron (MLP) has been compared in order to predict the DPN. The DPN data obtained from different hospitals. With the help of the Google-colab environment, the ML techniques SVM, Naïve Bayes, K-Nearest Neighbor (KNN) and DL technique MLP were implemented using the Python programming language. The model has a higher accuracy of 80.07%, while using the DL MLP method. It was also compared to the other ML strategies SVM obtained with 77.0%, Naïve Bayes acquiring 72.06% and yielding 69.2% using KNN. The DL and MLA comparative study of DPN forecasting shows that DL-MLP provides faster and higher DPN diagnostic performance.