Open Access

ARTICLE

A Hybrid Deep Learning Model for COVID-19 Prediction and Current Status of Clinical Trials Worldwide

by Shwet Ketu^*, Pramod Kumar Mishra

Department of Computer Science, Institute of Science, Banaras Hindu University, Varanasi, India

* Corresponding Author: Shwet Ketu. Email:

(This article belongs to the Special Issue: Machine Learning and Computational Methods for COVID-19 Disease Detection and Prediction)

Computers, Materials & Continua 2021, 66(2), 1896-1919. https://doi.org/10.32604/cmc.2020.012423

Received 30 June 2020; Accepted 05 October 2020; Issue published 26 November 2020

Abstract

Infections or virus-based diseases are a significant threat to human societies and could affect the whole world within a very short time-span. Corona Virus Disease-2019 (COVID-19), also known as novel coronavirus or SARS-CoV-2 (Severe Acute Respiratory Syndrome-Coronavirus-2), is a respiratory based touch contiguous disease. The catastrophic situation resulting from the COVID-19 pandemic posed a serious threat to societies globally. The whole world is making tremendous efforts to combat this life-threatening disease. For taking remedial action and planning preventive measures on time, there is an urgent need for efficient prediction models to confront the COVID-19 outbreak. A deep learning-based ARIMA-LSTM hybrid model is proposed in this article for predicting the COVID-19 outbreak by utilizing real-time information from the WHO’s daily bulletin report as well as provides information regarding clinical trials across the world. To evaluate the suitability and performance of our proposed model compared to other well-established prediction models, an experimental study has been performed. To estimate the prediction results, the three performance measures, i.e., Root Mean Square Error (RMSE), Coefficient of determination (R2 Score), and Mean Absolute Percentage Error (MAPE) have been employed. The prediction results of fifty countries substantiated the fact that the proposed ARIMA-LSTM hybrid model performs very well as compared to other models. The proposed model archives the lowest RMSE, lowest MAPE, and highest R2 Score throughout the testing, under varied selection criteria (country-wise). This article aims to contribute a deep learning-based solution for the well-being of livings and to provide the current status of clinical trials across the globe.

---

Keywords

---

Citations