Estimación de daño de miocardio producido por el mal de chagas mediante técnicas no invasivas

Abstract

Chagas' disease affects about 10 million people in Latin America. It is mainly transmitted by the faeces of triatomine bugs. The illness in humans can be recognized with a conventional Machado-Guerreiro test. However, it evolves in different stages. To determine the miocardial damage inflicted by the disease in the chronic stage it is necessary to perform several expensive, time consuming and sometimes even inva- sive tests. In this article, a machine learning system (Support Vector Machines) is trained to determine the degree of damage using exclusively the cardiac signals obtained from High Resolution Electrocardiogram (HRECG). The final classifiers consist of two simple formulas whose implementation can be easily carried out without the need of any knowledge in computer sciences. The research provides three significant contributions in the subject. First, it attains high classification rates. Second, it provides the final solution in two simple equations. Finally, it implements an exhaustive method useful to determine the best set of QRS features to train the classifiers. © 2013 Springer.