Diagnosis of STEMI and Non-STEMI Heart Attack using Nature-inspired Swarm Intelligence and Deep Learning Techniques

Acute heart attack is associated with 30% mortality rate, among those 50% of death occur before arriving to a hospital. The ST elevated myocardial infarction (STEMI) and non-STEMI heart condition may lead to heart attack which can be prevented if detected ahead of time. 2D convolutional neural network (CNN) uses 2D data has been successfully applied to machine vision, plant disease diagnosis and medical field. Acquiring 2D images such as CT, MRI, and PET data can be prohibitively expensive. On the other hand, there are many 1D biomedical signals, such as ECG, that is more affordable and can be used for medical diagnosis of heart diseases as an example. The purpose of this paper is to propose the use of 1D CNN for medical diagnosis relying on more affordable 1D biomedical signal. To reduce the computational burden and enhance performance, the firefly algorithm (FA) is first applied to reduce the number of features needed for classification by the 1D CNN. The proposed 1D CNN combined with FA technique was applied to STEMI and Non-STEMI heart attack diagnosis using ECG signals. The method was trained and tested using A clinically available synchronously acquired ECG signal database from physionet was used to train and evaluate the performance of the proposed technique. The correctly classified outcome using FACNN is 84.84% with kappa statistics of .693, while average performance from other popular machine learning algorithm were 78.54% correctly classified outcome and kappa statistics of .56.