Portal de Periódicos da CAPES

Improved Adaboost Algorithm with Regression Imputation for Prediction of Chronic Type 2 Diabetes Mellitus

2021; Springer International Publishing; Linguagem: Inglês

10.1007/978-981-16-1089-9_54

2367-3370

M. Dhilsath Fathima, S. Justin Samuel,

Machine Learning and Data Classification

Chronic type 2 diabetes mellitus is a type of diabetes which causes high blood glucose level in the human. Pancreatic insufficiency and elevated sugar level in blood are the conditions of getting diabetes mellitus. Type 2 diabetes mellitus (T2DM) affects the human metabolism such that a person with T2DM does not responding to the insulin released by the body, so glucose is not going into a cell in a normal and generates insulin resistance syndrome or metabolic syndrome. The consequences of diabetics are diabetic retinopathy, neuropathy, kidney damage, heart disease, slow healing and skin diseases. Hence, attention must be given to saving a person with T2DM by controlling the complications of T2DM. A predictive model is required to predict the T2DM effectively to reduce the severity of T2DM outcome. The main objective of this proposed work is to develop a T2DM prediction model using machine learning algorithms which act as a decision-making system to predict the type 2 diabetes in person. This model uses improved adaptive boosting algorithm (iABA) to develop the diabetic predictive model. This iABA model utilized Pima Indians diabetes dataset for building predictive model. Pima dataset contains many missing values which are imputed using regression imputation method. To examine the effectiveness of the proposed predictive model, the classifier performance measures are used. The iABA classifier outcomes are compared with a typical machine learning models. The output of the performance metric shows that an improved Adaboost algorithm achieves high accuracy of 78.3% than other machine learning classifiers. This model could be used to assists with medical professionals to make prediction of type 2 diabetes and used to classify a person as a diabetic person and non-diabetic person.