Predicting Diabetes Mellitus With Machine Learning Techniques

Abstract

Diabetes mellitus is a chronic disease characterized by hyperglycemia. It may cause many complications. According to the growing morbidity in recent years, in 2040, the world's diabetic patients will reach 642 million, which means that one of the ten adults in the future is suffering from diabetes. There is no doubt that this alarming figure needs great attention. With the rapid development of machine learning, machine learning has been applied to many aspects of medical health. In this study, we used decision tree, random forest and neural network to predict diabetes mellitus. The dataset is the hospital physical examination data in Luzhou, China. It contains 14 attributes. In this study, five-fold cross validation was used to examine the models. In order to verity the universal applicability of the methods, we chose some methods that have the better performance to conduct independent test experiments. We randomly selected 68994 healthy people and diabetic patients' data, respectively as training set. Due to the data unbalance, we randomly extracted 5 times data. And the result is the average of these five experiments. In this study, we used principal component analysis (PCA) and minimum redundancy maximum relevance (mRMR) to reduce the dimensionality. The results showed that prediction with random forest could reach the highest accuracy (ACC = 0.8084) when all the attributes were used.

Keywords: decision tree; diabetes mellitus; feature ranking; machine learning; neural network; random forest.

FIGURE 1

The structural of two–layer-feed-back network in MATLAB. This figure is from MATLAB, which can describe this network working principle preferably. Where, W is representation the weight and b is the bias variable.

FIGURE 2

Decision tree structure by using all features and Luzhou dataset. In this figure, we can find the fasting blood sugar is an important index for predicting diabetes And weight, age also have the higher information gain and play vital roles in this method.

FIGURE 3

Decision tree structure by using all features and Pima Indians dataset. From this figure, we can find in this method glucose as the root node, which can indicate the index has the highest information gain and insulin and age play important roles in this method.

FIGURE 4

The results of using Luzhou dataset. According to this figure, we found the method, which used all features and random forest has the greatest performance. And the methods without blood glucose are not good.

FIGURE 5

The results of using Pima Indians dataset. From the figure, mRMR is friendly for this dataset and method only using glucose is not suitable for this dataset.