COVID-19: Recent Prediction Modeling techniques using Machine Learning - A Survey

Authors

  • Kashinath G. Chaudhary, Sanjeev J. Wagh

Abstract

The first Human respiratory coronavirus recognized in the 1960s by Tyrrell and co-researchers. Their biological studiesprovide information and characterization of corona virus pathogenic agents. For ongoing SARS Cov-2 (COVID-19) research, researchersin medicine field are using serologic techniques and technical experts are using various predictions & forecast models to gather maximum information of epidemiology of the human respiratory coronaviruses.

Prediction and Forecast Modeling plays a vital role in decision-making during the emergence situations of disease spread.  To know the global impact of COVID-19, we need accurate forecasting, widespread population information, confirmed cases and analysis on recoveries and deaths. Due to uncertainty in the computed estimates and hypothetical intervention of authorities, the actions on emerging pathogen always get delay.  The modeling for pandemic gives information regarding the early stages of disease egression and helps to optimal decision control and management, also highlights uncertainty in specific areas to take immediate remedial actions.

 In this paper the in-depth survey on Machine learning based forecast and prediction models are focused which is based on ongoing recent literatures. The parameters like the contributed system, dataset of city/country, modeling scheme used with their purpose, limitations of model and overall impression are explored. The outcome of the survey is every model has its own advantages and disadvantages and also implemented on variety datasets with different locations. The Prediction model suitability is based on accurate datasets, intervention situation and locations. There is no idealistic or perfect predation model which will fits for universal cases.

Published

2020-03-31

Issue

Section

Articles