Binary classification for heart disease using classical and new non-invasive features

Raydonal Ospina, Adenice G.O. Ferreira, Hélio M. de Oliveira, Víctor Leiva

---

Abstract

In this paper, we find features of heart disease in the context of machine learning to classify cardiac patients. We introduce new non-invasive features in the classification models for heart disease predic- tion. We revisit a heart disease data set available from the UCI machine learning repository. The results show that the new relevant features are the mean arterial pressure, pulsatile blood pressure index, and the resistance-compliance that can be used to improve the accuracy of machine learning algorithms to binary classification of heart disease