Heart Disease early detection using the Machine Learning

Abstract

This research work have objective to compare the accuracy of seven classification algorithms for a set of data on heart diseases, allowing early detection in the patient who will have suspicion or not of the presence of this disease that simply cannot be detected. detect . The scope of the project is limited to analyzing two hundred and ninety-nine patients who underwent their tests for the creation of this data set, with the possibility of continuing to carry out constant tests to improve the quality of the information in the test. For the classification process, 25% will be tested for training versus 75% of the records as a sample, which will go through each of the models using the Confusion Matrix and the Auc Roc curve. It was observed that the precision in the Linear Regression model was 87.3%, while in the Auc Roc curve a reliability of 91% was obtained for the MultiLayer Perceptron Prediction model. The information from the tests carried out allows obtaining considerable precision in terms of the Linear Regression model, indicating that it works correctly with it, however, with a greater number of records, the results currently obtained could be further improved.