Construction and research of a heart disease prediction model based on ensemble learning and deep learning

Heart disease, also known as cardiovascular disease, encompasses various types such as rheumatic heart disease, congenital heart disease, coronary heart disease, myocarditis, and more. For many years, heart disease has been one of the leading causes of death worldwide, and its occurrence is closely related to various risk factors. Currently, the diagnosis of heart disease is mainly made by healthcare professionals, but its drawback lies in being heavily influenced by the doctor's subjectivity and having a slow diagnosis process. Therefore, there is an urgent need for an accurate and fast method to assist healthcare professionals in making diagnoses, improving diagnostic accuracy, and reducing the burden on healthcare workers. This paper first establishes four heart disease prediction models—Naive Bayes, Logistic Regression, Decision Tree, and KNN—using various indicators of patients, and then studies and compares the performance of these machine learning models. Furthermore, the idea of ensemble learning is introduced, using the above four types of machine learning models as classifiers to build ensemble learning hybrid models with Random Forest, LightGBM, and XGBoost. Experimental results show that the performance of the LightGBM model is better than that of any single machine learning model. Subsequently, various deep learning methods are used for feature learning of electrocardiogram (ECG) images. Testing results show that the EfficientNet model achieves a prediction accuracy of 0.928 and an AUC value of 0.934, demonstrating good model performance. By combining ensemble learning with deep learning methods, the model's prediction accuracy and reliability can be greatly improved. Additionally, this approach has advantages such as objectivity, low cost, and high efficiency.