CardioSmart Analyzer: A Machine-Learning-based ECG Diagnostic System for Early Detection and Interpretation of Cardiovascular Disease

Abstract

CardioSmart Analyzer is an intelligent framework that automatically evaluates 12-lead electrocardiogram (ECG) measurements to identify a wide range of cardiovascular diseases (CVDs). The study introduces a machine-learning prediction pipeline for recognizing cardiac abnormalities from ECG-derived data. Twelve-lead ECG records from 2,840 patients were collected at hospitals in Gazipur, Bangladesh, between June and August 2024. Expert assessments were used to annotate 68 separate cardiac diagnoses. A Random Forest (RF) classifier was trained to distinguish these conditions and was assessed alongside several benchmark machine-learning algorithms. The system attained 97.89% accuracy in binary classification and 95.43% accuracy in the 68-class multiclass task. These outcomes suggest that the framework can identify cardiovascular irregularities from ECG information with high reliability. Its overall performance indicates that CardioSmart Analyzer may serve as a clinical decision-support application for early screening and diagnosis, reduce interpretation workload, and improve the efficiency of patient-care decisions.