A Novel Advanced Performance Ensemble-Based Model (APEM) Framework: A Case Study on Diabetes Prediction

The use of machine learning algorithms for detection and prediction of diabetes diseases has been used widely in many studies. However, the accuracy of the prediction results in existing studies remained low. This research proposed a novel Advanced Performance Ensemble-based Model (APEM) framework as an enhancement method designed to enhance the accuracy level of predicting diabetes concerns compared with previous studies. Three main contributions are offered by APEM in this study as novel contributions. First, the paper discusses how to select the most appropriate preprocessing methods for the data, second, how to select and experiment with a number of machine learning algorithms as part of the ensemble learning process, and thirdly, how to Achieve the highest accuracy value compared to existing research. In general, there are three main stages in the APEM framework, the first stage is preprocessing data, the second stage is the ensemble method that uses five different machine learning algorithms, and the third stage is the second layer of the ensemble method with one machine learning algorithm. The result of this research produces better prediction results of diabetes prediction with an improvement in accuracy value of 99.06% compared with previous research, with a note that both this study and previous research utilized the same Pima Indian dataset and machine learning approach for their prediction.