Literature Reviews: Penggunaan Multilayer Perceptron untuk Klasifikasi Diabetes Melitus

Abstract

Diabetes Mellitus or commonly known as diabetes is a chronic metabolic disease characterized by high blood sugar levels (hyperglycemia) due to impaired body ability to produce insulin. Based on data from the International Diabetes Federation (IDF), the number of diabetes sufferers will increase rapidly in 2024 to 700 million people. Therefore, we need to find out what are the triggers for contracting diabetes mellitus. One of them is by using the machine learning method. Machine learning is used to classify what factors can be the possibility of contracting diabetes mellitus. One of the well-known methods for making this classification is the Multilayer Perceptron (MLP) method, which is a type of artificial neural network (ANN) consisting of several layers, where each layer has nodes that are interconnected. Its advantage is that it is able to handle non-linear relationships between complex data features - including patient data and diseases in patients - so this method is said to be very relevant for this study. Researchers also compared the accuracy of MLP with several other algorithms, such as Random Forest, Support Vector Machine, and K-Nearest Neighbors. This is intended to evaluate the effectiveness of MLP in classifying diabetes mellitus compared to other methods. In addition, researchers also want to overcome the weaknesses of traditional methods in classifying diabetes and offer solutions based on artificial intelligence, by utilizing MLP in processing medical data and paying attention to parameters or features that can affect patients with diabetes. Several techniques in machine learning, such as regularization and hyperparameter optimization can prevent overfitting, data normalization and dimensionality reduction are used to improve the quality of input given to the model, so as to maximize accuracy and make the diagnosis process faster and more precise. The results obtained show that MLP has good performance in classifying this disease, compared to other algorithms. MLP gets more stable and higher results. Overall, it can be said that the application of MLP makes a significant contribution to improving the diabetes diagnosis system and is expected to be applied in the health system.