ABSTRACT
The COVID-19 pandemic is still wreaking havoc on health care infrastructure, the economy, and agriculture. This pandemic impacted masses at physical, mental and financial levels. Several of the fastest-growing economies in the world are struggling as a result of the epidemic’s intensity and contagiousness. Forecasting the number of infected COVID-19 patients could be helpful in preparing future hospital resources and planning due to the expanding diversity of cases and the resulting load on healthcare providers and the government.
Due to the complexity of virus propagation, even the most well-known computational and mathematical models have been shown to be incorrect. This paper focuses on developing and implementing artificial intelligence (AI) algorithms to predict COVID-19 propagation using available time-series data. We have focused on 10 hotspot countries (China, Germany, Iran, Italy, Spain, United States, France, Turkey, United Kingdom and India). To forecast COVID-19 confirmed, active, recovered and deaths, we used a suggested model based on long short-term memory (LSTM) which has an input layer that is followed by three LSTM layers and ten hidden units (neurons).
