Heart disease prediction with big data has become an interesting study topic when observing intelligent healthcare applications. Big data analytics for heart disease prediction must be accurate and lightweight. Deep Learning approaches over the past few years recently performed accurately and lightweight covering numerous domains. Despite that, applying deep learning with big data for heart disease prediction is a demanding research problem. Due to the class imbalance and dimensionality nature, accurate and precise big data analytics for heart disease prediction remains a major issue to be addressed. Hence, the proposed method’s main objective is to boost big data analytics for heart disease prediction efficiency using a hybrid feature engineering method, Weighted Stouffer Quantum Mutual, and Tangent Radial Deep Learning-based heart disease prediction (WSQM-TRDL). Weighted Stouffer and Quantum Mutual Graph-based Feature Filtering are initially applied to the raw dataset for obtaining class-balanced dimensionality-reduced features. Next, we design a hybrid model to efficiently classify big data using the Tangent Kernel function and Radial Deep Learning. For accurate and precise heart disease prediction the hybrid learned features are fed successfully into the deep learning classifier Radial Basis Function and also planned to conduct simulation using Python to measure and validate the performance metrics in terms of Accuracy, Error rate, Prediction time, and sensitivity.