Research on predicting heart attack through active learning

Machine learning has been widely used in medical diagnosis areas, such as detecting disease at its earliest stages, helping doctors diagnose patients more accurately, and providing advice for patients. Active learning is a new machine learning method. Its main goal is to find high-information samples from many data sets to reduce the number of training sample sets, reduce computational complexity, and improve the generalization ability of the classifier. This paper describes how to create a model derived from machine learning techniques and use a pool-based active learning algorithm to analyze and predict whether a patient is at risk of heart disease. The author uses principal component analysis (PCA) to create a framework to extract features and calculate a mathematical model to select relevant features using relevant constraints. This experiment combines two datasets that are analyzed commonly in one dataset. The classification accuracy of this dataset was obtained by k-fold cross-validation. The experiment compared three modeling options, which are Random forest (RF), K Nearest Neighbors (KNN), and Multilayer Perceptron (MLP). The best model tested in the experiment is RF, which achieved an average accuracy of 88%.