Enhancing the Predictive Performance of Heart Disease Detection on The Cardiovascular Disease Dataset Using Gated Recurrent Unit and Light Gradient Boosting Machine

Abstract

Heart disease is one of the leading causes of mortality worldwide, making early and accurate diagnosis essential for improving patient survival rates and reducing healthcare costs. Traditional diagnostic techniques often face challenges in achieving high predictive accuracy due to the complexity and variability of cardiovascular data. To address these limitations, this research proposes an intelligent heart disease prediction framework using two advanced approaches: Gated Recurrent Unit (GRU) and Light Gradient Boosting Machine (LightGBM). The proposed models are implemented and evaluated on the Cardiovascular Disease Dataset to enhance predictive performance and classification efficiency. Initially, the dataset undergoes preprocessing techniques including data cleaning, normalization, and feature preparation to improve data quality and model performance. The GRU model is employed to capture complex hidden patterns and temporal relationships within the dataset, while the LightGBM model is utilized for efficient gradient boosting-based classification with reduced computational complexity. The performance of both models is evaluated using Accuracy, Precision, Recall (Sensitivity), Specificity, and F1-Score. Experimental results demonstrate that the proposed models significantly improve heart disease prediction performance compared to conventional machine learning techniques. Among the evaluated models, LightGBM achieves superior classification efficiency and faster training performance, whereas GRU effectively learns complex feature representations from medical data.