Machine Learning-based Prediction of Diabetes for Improved Healthcare

Diabetes is a disease that can lead to severe tissue damage and dysfunction, and to improve the accuracy of one's prediction of early diabetes, patient datasets can be used to build Machine Learning (ML) and Deep Learning (DL) models to make the results more accurate and valid. There have been impressive advances in the integration of Artificial Intelligence (AI) and Machine Learning techniques in healthcare systems. This paper presents a comparative analysis of machine learning and deep learning algorithms for diabetes. The dataset used in the experiment is available at www.kaggle.com. In our experiments, we compared and analyzed the classification accuracies of each dataset under different classification algorithms and compared and analyzed the results with the accuracies of the corresponding algorithms listed in the references. The results show that in most cases the proposed algorithm outperforms the references in terms of classification accuracy, and the difference in this result is due to different data preprocessing. The original dataset will be further improved in the data preprocessing section and feature engineering will be further investigated at a later stage. Preprocessing the data and adjusting the model parameters can lead to better classification results. The accuracy of each model varies, and by comparing the results of the various algorithms, it is found that the random forest algorithm and the multilayer perceptron (MLP) algorithm have better accuracy than the other methods, and this finding lays the foundation for subsequent related research.