A novel research on a heart disease prediction model based on hybrid intelligent optimization of deep networks

As one of the leading causes of mortality worldwide, the early and precise prediction of heart disease is crucial for effective clinical intervention. This research proposes a novel cardiac prediction framework, which integrates a bidirectional temporal convolutional network (BiTCN) and a bidirectional gated recurrent unit (BiGRU), to construct a feature fusion model. This architecture effectively captures the spatiotemporal dependencies within the cardiac data. To address the challenge of optimizing the model hyperparameters, this research innovatively introduces a hybrid strategy combining Bayesian optimization with the Arctic Lemming Algorithm (ALA), which significantly enhances the model’s generalization capability. The simulation results demonstrate that, the proposed intelligent hybrid optimization strategy effectively addresses the challenges of high-dimensional modeling of medical data, providing reliable intelligent decision support for the early clinical diagnosis of heart disease.