A Machine Learning Analysis on Diabetes data and Integration of Data securities through Euclidean Encryption

Abstract

The improvement in living standards as in case of diabetes is becoming more prevalent. This begs the issue of how to properly diagnose diabetes in the first place. Diagnosis of diabetes is accomplished via the use of fasting, glucose, and random blood glucose levels. The sooner it is detected and treated, the more manageable it is. Machine learning may be used to assist individuals in evaluating diabetes mellitus utilizing data from their daily physical examinations as a doctor's reference. Valid features and classifiers are two of the most difficult issues in machine learning to solve. Several methods, including machine-learning approaches such as SVM, DT, and logistic regression, have recently been developed for the prediction of diabetes. Diabetes is differentiated from non-diabetics, and the distinction is made using a nervous inference. Automated machine learning (ML) is extensively utilized in diabetes prediction, and it produces more accurate findings. A decision tree is a medical machine learning technique that is widely used. Random forests are generated by a large number of decision trees. Machine learning performance has been improved in a variety of ways as a result of the usage of neural networks. When it comes to predicting diabetes, this research utilized decision trees, LR, which had higher accuracy when compared to other methods. Consumers' diabetes data may be gathered and recovered using encryption and the Euclidean Algorithm, which may be used to prevent this from happening. It is this algorithm that provides the security feature as well as making it simpler for diabetics to use Internet of Things devices to manage their condition. This paper explores the deep investigation of Diabetic prediction though data as well protect the user data though encryption.