Enhancing the Efficiency of Diabetes Prediction through Training and Classification using PCA and LR Model

Abstract

In this paper, we introduce a new approach for predicting the risk of diabetes using a combination of Principal Component Analysis (PCA) and Logistic Regression (LR). Our method offers a unique solution that could lead to more accurate and efficient predictions of diabetes risk. To develop an effective model for predicting diabetes, it is important to consider various clinical and demographic factors contributing to the disease's development. This approach typically involves training the model on a large dataset that includes these factors. By doing so, we can better understand how different characteristics can impact the development of diabetes and create more accurate predictions for individuals at risk. The PCA method is employed to reduce the dataset's dimensions and augment the model's computational efficacy. The LR model then classifies patients into diabetic or non-diabetic groups. Accuracy, precision, recall, the F1-score, and the area under the ROC curve (AUC) are only a few of the indicators used to evaluate the performance of the proposed model. Pima Indian Diabetes Data (PIDD) is used to evaluate the model, and the results demonstrate a significant improvement over the state-of-the-art methods. The proposed model presents an efficient and effective method for predicting diabetes risk that may have significant implications for improving healthcare outcomes and reducing healthcare costs. The proposed PCA-LR model outperforms other algorithms, such as SVM and RF, especially in terms of accuracy, while optimizing computational complexity. This approach can potentially provide a practical and efficient solution for large-scale diabetes screening programs.