An efficient covid-19 prediction using Penguin Pelican optimization-based recurrent dropout-enabled hybrid deep CNN-BILSTM classifier

Problem statementThe global COVID-19 pandemic has had devastating consequences, emphasizing the need for swift and reliable coronavirus patient detection to enhance treatment and reduce transmission.MethodologyThe extracted features are fed into a hybrid deep classifier that combines Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) layers. The recurrent dropout technique is employed within the BiLSTM layer as a regularization method to prevent overfitting. This architecture leverages the strength of CNNs in feature extraction and the sequential memory handling of BiLSTMs. The research introduces a specialized optimization algorithm, the Penguin Pelican Optimizer (PPO), which is a fusion of characteristics from the Emperor penguin optimizer and the Pelican Optimization algorithm. This algorithm is designed to fine-tune and enhance the efficiency of the system by optimizing model parameters, weights, and biases.ResultsThe findings of this study are highly encouraging. The proposed Penguin Pelican Optimization-based recurrent dropout-enabled hybrid deep CNN-BILSTM approach achieved remarkable results, with accuracy rates of 95.57%, 96.08%, and 95.54% for training percentage, and 95.75%, 96.28%, and 95.70% for k-fold. These outcomes surpass the performance of existing methods. This research underscores the practicality and efficiency of the developed methodology for COVID-19 prediction. It offers valuable insights that can be instrumental in ensuring timely and accurate patient management, thus contributing significantly to the global effort to combat the pandemic.ConclusionThis research underscores the effectiveness and efficiency of the developed approach for COVID-19 prediction, providing valuable insights for improved patient management in the battle against the pandemic.