Using Machine Learning In Predicting Diabetes

Abstract- Diabetes mellitus, commonly known as diabetes, is a metabolic disease that causes high blood sugar. We all know that the hormone normally passing through the blood to our cells for gathered energy. Any Diabetes contains unprocessed high blood that can damage our nerves, eyes, kidneys, hearts, and other organs. In certain times, there are a lot of people suffer or affected by diabetes for the reason the people have to go to the diagnostic center, hospital, or clinic for tests. So naturally, the management has to store the required tests report and provide a proper diagnosis based on them report. But the rise in machine learning approaches solves this critical problem. ML (Machine Learning), DS(Data Science) and AI (Artificial Intelligence) play a momentous role in hospitals, diagnostic, clinic, or any Healthcare industries. This type of place must behave a lot of large volume databases. Using a Data analysis technique, one can study a huge dataset and find deeper information, deeper patterns, deeper symptoms from the data, and predict outcome accordingly. The intension of this recitation is to scheme a unique model that can predict diabetes with maximum accuracy. In the existing method, the classification and prediction accuracy is high. In this paper, I have proposed a new diabetes prediction model that model I already deployed on the server. For getting the best diabetes prediction I include some extra inputs along with regular inputs such as Age, Glucose, BMI, etc. There are six classification of ML (machine learning) algorithms such as Gradient Boosting Classifier, Logistic Regression, Decision Tree, SVM, K-Nearest neighbors, and Naïve Bayes are used in this experiment to detect diabetes at an early stage. n this case study I used Pima Indian Diabetes Database (PIDD) which is collected from the UCI machine learning repository. The accuracy of all the six algorithms is measured on different techniques like Precision, Accuracy, F-Measure, and Recall. From this study, I got the highest accuracy of 83.76% from Gradient Booster Classification comparatively than any other algorithms using in this experiment. These Accuracies are corroborated apply Receiver ROC (Operating Characteristic Curve) in an appropriate way.