AI-Driven Discovery of Novel Biomarkers for Early Cardiovascular Disease Detection

Abstract

Cardiovascular diseases (CVDs) are the main cause of mortality in the world, and it can develop without symptoms until the late stage when the damage becomes irreversible. The traditional risk assessment tools, which are based on linear modeling and constrained clinical characteristics, do not have the sensitivity of early identification. This paper introduces an artificial intelligence based system of identifying new biomarkers that can be used to predict CVD early. The model with the help of ensemble-based machine learning, i.e. Random Forest, incorporates systematic data preprocessing, complex feature selection and explainable artificial intelligence (XAI) methods to determine clinically meaningful biomarkers. The model was tested with extensive measures of evaluation, cross-validation and external data showing a high prediction rate, stability, and readability. Identified key biomarkers such as age, systolic blood pressure, cholesterol, glucose levels, and body mass index are consistent with the known clinical information but represent non-linear, complex interactions. The model can help patients by promoting individual risk stratification and improving clinical trust by adding SHAP-based interpretability. This study highlights the possibility of AI to revolutionize early cardiovascular diagnostics to provide a scalable, transparent, reliable, and understandable instrument of preventative healthcare.

DOI: https://doi.org/10.66669/WT.v35iS2.287