An Intelligent Prediction Model of COVID-19 in India using Hybrid Epidemic Model Manoj Kumar Department of Computer Science & Engineering Delhi Technological University Delhi, India mkumarg@dce.ac.in Jatin Bareja Manufacturing process and automation engineering Netaji Subhas Institute of Technology Delhi, India jatinb.mp@nsit.net.in Manjot Singh Manufacturing process and automation engineering Netaji Subhas Institute of Technology Delhi, India manjots.mp@nsit.net.in Rupanshu Sharma Manufacturing process and automation engineering Netaji Subhas Institute of Technology Delhi, India rupanshus.mp@nsit.net.in Abstract—This article presents the current situation of COVID 19 spread in India and how it is impacted by various measures taken by the administration. Data source is taken (till 15th June 2020) from World Health Organization (WHO) to study various trends and pattern. Hybrid epidemic susceptible-infected recovered model is used to make predictions at every stage of coronavirus. The basic reproduction number R0, is constructed using a logistic function. An improved or hybrid epidemic model is build based on various other factors to build a logistic infection rate to analyse the patterns and trend in the data. Secondly, ICU beds and ventilators available per 100,000 inhabitants is also taken in account for critical. Also, the timeline for changing the stage of coronavirus is used from various reliable resources. Moreover, in comparison with the traditional SIR models, accuracy in the predictions has significantly increased with the R 2 value of 96.8 percent for the future days in India. Index Terms—Coronavirus (COVID-19) prediction, epidemi- ological models, hybrid epidemic model, least square fitting, logistic R0 I. I NTRODUCTION A newly identified coronavirus, SARS-CoV-2, which has caused a worldwide pandemic of respiratory illness, also known as COVID-19. Currently, researchers are looking into the virus and has figured out that the virus spreads through the drops that is released into the air through infected person. According to the recent data (18th July 2020) by John Hopkins University [1], currently there are around 13 million people who got infected by the Corona Virus all around the world. India is currently second most populated country in the world with one of the highest GDP and growing very rapidly in terms of economy [10]. Therefore, this study focus on analyzing the results which is based on different steps taken by the Indian government during the epidemic. First case of coronavirus in India was reported on January 30, 2020 and the cases are rapidly increasing since then. In response, since 22nd March there has been a lockdown in the whole country and every activity is happening with the permission of administra- tion unit of India and all the international travels have been either banned or monitored closely. With the development of the pandemic in India, it becomes important to understand the current situation and its impact through statistical analysis methods and prepare for ahead which should be helpful for the Indian government and medical practitioners. Whenever there is outbreak of such type of diseases having large scale implications, people try to build various epidemic models and try to understand and predict the trend and future implications to take the appropriate control measures. Most frequently used models for such cases are SI (susceptible – infected), SIR (susceptible – infected – recovered), and SEIR (susceptible – exposed – infected – recovered) models [11],[2] . In these models stages such as susceptible, exposed, infected, recov- ered are interconnected with each other through differential equations and due to their capabilities of analyzing trend and development in data, they are used to make future predictions for diseases like COVID 19 [3] , [22]. However, according to these models, everyone who tested positive for coronavirus would have the same rate of infection. From the results of these model, a mere overview of the trend has been obtained and, thus, it have certain limitations. The accuracy of the data reported plays a major role in analyzing trends of the disease but data alone is not enough, the way administration handles the situation must be consider in the model for results to be accurate. Agrawal [21], in their study, they used a particular rate function which was not linear in nature. An epidemic model (SIR) with treatment function and rate of infection was de- scribed by [9]. In their study they focused on stability and set the dynamics of the model accordingly to build a basic reproduction number R0. It was represented that the model they used own a reverse bifurcation. It was seen that they used geometry to understand the equilibrium with global asymp- totic stability. Adebimpe [1] portrayed, equilibrium that is disease free is locally asymptotically stable if their production number R is less than one and equilibrium that is endemic Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020) IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9 © IEEE 2020. This article is free to access and download, along with rights for full text and data mining, re-use and analysis 389