Article Preview
Top1. Introduction
Currently, more than one million people around the world face the severe consequences of the outbreak of the novel Coronavirus 2019 (nCov). The first case of human infection by a nCov or Wuhan virus or 2019- nCov was reported in Wuhan, China1. The greatest challenge of this infectious disease is the human-to-human transition of nCov that would rise up the infected cases exponentially. On 30 January 2020, World Health Organization (WHO) issued a worldwide health emergency warning notice2, designating that 2019-nCoV is of urgent global concern. The morbidity and mortality rates for the infection of 2019-nCoV are uncertain at the early stage (Sparrow 2020) especially for young ones and aged people. WHO has estimated the reproduction factor (R0) of nCov is 2.7. In order to control the wide and quick spread of the nCov, public health sectors took reliable preventative measures and imposed curfew or lockdown infested cities in China, United States, India, and other countries also. This is to limit the social distance among people and to avoid the transmission of this novel virus via humans to humans.
Over the past decade, machine learning techniques have gained momentum and have played an important role in many domains of research. Machine Learning (ML) is a subset of artificial intelligence that optimizes data through a series of mechanisms and provides novel insights or form of data to take timely active or preventive measures. In particular, it has a tremendous impact on data analysis and data science. It better understands the data and its processes, makes predictions about the future based on historical data / empirical data, and automatically classifies a group of data called classification. There are two types of learning techniques in ML techniques: unsupervised and unsupervised learning techniques.
Supervised learning techniques (regression, classification, and regression trees (CART) and nave Bayes) use labeled data to train input and output known mechanisms. Unsupervised learning techniques (Association Rules and Clustering.) use unlabeled data, inputting raw data directly to these methods without knowing the output of that data.
Machine learning techniques also can be used to develop standard mortality models. Deprez et al. (2017), used machine learning algorithms to fit and assess the mortality model. The regression approach used to detect the weaknesses of different mortality models by the authors. In Hainaut (2018), artificial neural networks (ANNs) used to find the latent factors of mortality and forecast them. Richman and Wüthrich (2018) extended the Lee-Carter model to multiple populations using neural networks.
Generally, Gaussian process models have been widely used in engineering based optimization applications (Razavi et al. 2012). In Raghavendra and Deka Raghavendra and Deka (2016), a combination of GPR and adaptive neuro-fuzzy inference system (ANFIS) used to forecast the ground- water level. In Roy and Datta (2018), an extensive comparative study was carried out between several surrogate models, comprising GPR, using simulation-optimization methodology with uncertainty pa- rameters. At the end, they had concluded that the GPR models and their ensemble were efficient methods with respect to prediction accuracy.
In this work, the Gaussian process regression model with the optimal hyperparameter is used to develop COVID-19 mortality models for five different countries (USA, Italy, Japan, Germany, and India). Also, evaluate the effectiveness of these models in the model evaluation and prediction of COVID-19 mortality rates for India. The purpose of this study is mortality prediction, which uses machine learning techniques, to clearly identify patterns that cannot be identified with the standard mortality model. The rest of this paper is plotted as follows. Section 2 describes the proposed method for predicting COVID-19 mortality for India with an early-stage time-series dataset. The results and discussion of the empirical study are presented in Section 3. Section 4 concludes this work with an extension of possible future work.