Improved Coefficient Based Chicken Swarm Optimization With Exponential Distribution Based Long Short-Term Memory For Covid-19 Prediction

Abstract

The rapidly spreading COVID-19 infection created an intensive impact on physical as well as psychological wellbeing of the public throughout the world. This pandemic has created a challenging situation in the healthcare system around the world. It created a huge demand for hospital care and instruments. Early identification of the infections with the help of efficient screening as well as diagnostic procedures can help in reducing the demand on health care. Various Prediction models combining number of characteristics for estimating the risks of infection have been created. The previous system designed a Weighted Butterfly Optimization Algorithm (WBOA) with Intuitionistic fuzzy Gaussian function based Adaptive-Neuro Fuzzy Inference System (IFGF-ANFIS) classifier for covid-19 prediction. However, it has issue with classification results.  For solving this issue, the applied system designed an Improved Coefficient based Chicken Swarm Optimization (ICCSO) with Exponential Distribution Based Long Short-Term Memory (EDLSTM) approach for covid-19 prediction. Initially, the COVID-19 Patient dataset is taken as an input and normalized by using Z -score normalization. Then the dimension reduction is done by using Independent Component Analysis (ICA). For improving the classification accuracy, Improved Coefficient based Chicken Swarm Optimization (ICCSO) algorithm is used for optimal feature selection. Based on the selected features, an Exponential Distribution based Long Short-Term Memory (EDLSTM) is utilized for classifying the people having infection possibility. Experimental analysis was conducted on a real data set. Results show that the proposed system attains better performance compared to previous system in terms of accuracy, precision, recall and f-measure.