Computer-Aided Decision Support System for Diagnosis of Heart Diseases

Abstract

Background: Cardiovascular diseases (CVDs) are the leading causes of death worldwide and the number of people dying from these diseases is steadily increasing. The rapid economic transformation leading to environmental changes and unhealthy lifestyles increase the risk factors and incidence of cardiovascular disease. The limited access to health facilities, lack of expert cardiologists, and lack of regular health checkup trends make CVD a major cause of mortality in low-resource settings. Computer-aided diagnosis using artificial intelligence techniques (AI) can help reduce the mortality rate by providing decision support to experts allowing early diagnosis and treatment. Methods: In this paper, an AI-based computer-aided heart disease diagnosis decision support system has been proposed using clinical data, patient information, and electrocardiogram (ECG) data. The proposed system includes three modules: an ECG processor module that allows cardiologists to process and analyze the different waveforms, a machine learning-based heart disease prediction module based on patient information and clinical data, and a deep learning-based 18 heart conditions multiclass classification module using 12-lead ECG data. A user-friendly user interface has also been developed for ease of use of the proposed techniques. Results: The heart disease prediction module was found to be 100% accurate in predicting heart disease based on clinical and patient information, and the multiclass classification module was 93.27% accurate, on average, in classifying heart conditions based on a 12-lead ECG signal. The ECG processor also provides quick diagnosis by analyzing important ECG waveforms and segments. Conclusion: The proposed system may have the potential for facilitating heart disease diagnosis. The proposed method allows physicians to analyze and predict heart disease easily and early, based on the available resource, improving diagnosis accuracy and treatment planning.