Machine Learning Approach in Medical Diagnosis: Predicting Diabetes Complications

Nowadays, diabetes has become a worldwide disease that will have a bad impact on the human body’s health. Understanding body condition and detecting the signs of diabetes early is key to preventing diabetes from becoming serious. Therefore, an effective machine learning technology is implemented for predicting diabetes with different features from the diabetes dataset. This research is aimed to implement the machine learning model (Logistic Regression) to predict diabetes and identify the effect of parameters on accuracy. The real diabetes patient dataset with 390 records is from Data World and applied to the model. The parameter which is “penalty” is set with different values to test the accuracy of the model. 30% of data will be used for testing data and 70% of data as training data. The results of model accuracy are more than 90%. The model is implemented well to predict diabetes and further experiments for testing the model are needed to improve the accuracy of the model.