Astha
Vol. 16, Issue 1, Jul-Dec 2023
Abstract:
Heart disease is one of the leading causes of morbidity and mortality worldwide, posing a significant burden on healthcare systems globally. Accurate and early heart disease prediction is critical for effective clinical intervention and management. This paper investigates the potential of synergistically integrating machine learning (ML) and deep learning (DL) techniques to enhance heart disease prediction. Traditional ML techniques, such as RF and SVM, have been shown to work efficiently on structured clinical data. DL architectures, including CNNs and RNNs, perform well with unstructured data like medical images or time-series signals. The proposed hybrid framework combines the best of both approaches: the interpretability of ML and the feature extraction capability of DL. The research assesses the hybrid model using state-of-the-art models, advanced feature engineering techniques, and diverse clinical datasets. Its findings demonstrate that the hybrid framework performs significantly better accuracy, sensitivity, and specificity metrics than an individual ML or DL model. Moreover, integration can ensure a better identification of significant predictors without jeopardizing performance and interpretability. The paper concludes with insights into the clinical applicability of the hybrid model, identifying challenges and proposing recommendations for future research. This approach can potentially transform cardiovascular disease management and improve patient outcomes by advancing predictive capabilities. The study proposes a hybrid framework capable of improving prediction accuracy. The analysis incorporates state-of-the-art models, feature engineering techniques, and clinical datasets to identify risk factors and predict outcomes effectively. The findings suggest that the proposed hybrid approach surpasses individual ML and DL models in accuracy, sensitivity, and specificity. The paper concludes with future research directions and recommendations for clinical implementation.
DOI: http://doi.org/10.37648/ijrmst.v16i01.017