Hybrid feature selection with novel deep learning model for COVID-19 risk prediction

In the past five years, the COVID-19 epidemic has created unparalleled difficulties throughout the world. The COVID-19 virus has not entirely disappeared all over the world, and it is still spreading with new transformations and new variants throughout the globe. Hence, the requirement of exact and appropriate risk prediction can emphasize the care of patients and effectively allocate the resources for improving their lifespan. The prediction of risk factors for COVID-19 is highly needed for minimizing these hazards. As a result, this work develops the Fuzzy-Deep Kronecker Network + Deep Recurrent Neural Network (Fuzzy-DKRNN)-based Covid-19 risk prediction model. The data normalization is the primary stage of this process. The log scaling is utilized to provide the normalized data effectually. For selecting the significant features, the hybrid similarity measure is employed. Moreover, the Fisher score with Pearson’s Correlation Coefficient (PCC) is considered as the hybrid similarity measure. The proposed Fuzzy-DKRNN model effectively performs the COVID-19 risk prediction. In addition, the accuracy, precision, recall, and F1-Score are utilized to validate the Fuzzy-DKRNN with the ideal values like 0.908, 0.900, 0.918, and 0.890 are attained.