Quantum-inspired seagull optimised deep belief network approach for cardiovascular disease prediction.

The early detection and accurate diagnosis of cardiovascular diseases is vital to reduce global morbidity and death rates. In this work, the quantum-inspired seagull optimization algorithm (QISOA) combined with a deep belief network (DBN) is proposed to improve the identification of cardiovascular disorders. As part of preprocessing, cleaning, transformation, and standardization are performed to eliminate noise, inconsistencies, and scaling issues in the data. QISOA is used to optimize the weights and biases of the DBN model, enhancing its prediction efficiency. The algorithm incorporates quantum mechanics concepts to develop its exploration potential further, leading to faster convergence and increased global search efficiency. Optimized DBN provides efficient acquisition of hierarchical representations of the data, which results in improved feature learning and classification accuracy. The publicly accessible Cleveland Heart Disease dataset is used to assess the performance of the suggested model. Extensive experiments are conducted to demonstrate the superior performance of the QISOA-optimized DBN model compared to traditional machine learning and other metaheuristic-based models. Initially, machine learning models such as support vector machines, decision trees, Random Forests, multi-layer perceptrons, and fully connected networks were considered for comparison with the cardiovascular predictive performance of the DBN model. Further, meta-heuristic optimization algorithms such as particle swarm optimization, genetic algorithm, grey wolf optimization, cuckoo search optimization and crow search algorithm are combined with the machine learning models and the classification efficiency is evaluated. Additionally, few state-of-the-art techniques proposed in the existing literature are investigated and compared against the proposed model. It was evident from the comprehensive performance assessment of the proposed model that it yields a higher accuracy of 98.6% with precision, recall, and F1-scores of 97.6%, 96.8%, and 97.1%, respectively, compared to other traditional and existing models for cardiovascular disease prediction.