An AI-Aided Cardiovascular Care Unit for the Classification of Arrhythmias and Cardiac Maladies

Abstract

Pakistan, like many developing countries, faces several challenges in providing standard health care facilities, particularly in less privileged and least developed areas. Cardiovascular diseases (CVDs) are a leading contributor to the reported mortality rate in Pakistan, constituting approx imately 30-40 percent of all documented deaths. Unfortunately, the late response in providing specialized health emergency services exacerbates the problem. Cardiologists, who are respon sible for providing specialized care to CVD patients, spend a significant amount of time diag nosing the condition, leaving them with less time to focus on treatment, which can be a matter of life and death. To address this issue, an AI-aided system has been proposed that focuses on swift and accurate diagnosis of CVDs and Cardiac Arrhythmias. The system employs an en semble model that consists of a machine learning (ML) model and a deep learning (DL) model. The proposed ensemble model only takes non-invasive cardiac parameters as an input. The ML model assesses cardiac parameters such as temperature, blood pressure (BP), oxygen saturation (SpO2), and heart rate (HR), while the deep learning model analyzes electrocardiogram (ECG) signals. This combination of cardiac parameters and ECG analysis can provide accurate diagno sis and treatment recommendations in a specific context. To develop a complete cardiac dataset, the ECG dataset from MIT-BIH was extended by using another dataset consisting of temper ature, BP, SpO2, and HR. The proposed ensemble paradigm was evaluated by using various evaluation measures including accuracy, F1-score, recall, precision, specificity, and sensitivity. Our findings indicate that the proposed framework outperformed other cutting-edge models for the given cardiac dataset. Moreover, this research promises to predict maximum CVDs and Arrhythmia classes by applying smart AI techniques. Ultimately, the proposed AI-aided sys tem can significantly reduce the workload of cardiologists by enabling them to focus more on treatment rather than diagnosis.