IIT Kharagpur develops COVID-19 Predictive Model for Decision-making

Share this:

The study also predicts that new cases of the disease will continue until at least the end of September 2020.

India Today        Hindustan Times                 Times of India        Economic Times (Health)         Business Insider           News 18       Outlook          The New Indian Express          Down to Earth          NDTV          The Hindu            India Blooms             Telangana Today       Lokmat             Navbharat Times           Dainik Jagran

IIT Kharagpur has developed a prediction system to help predict the future spread of COVID-19 and help decision making in health-care, industry, economy, and even academics. Prof. Abhijit Das from the Dept. of Computer Science and Engineering has developed a logistic model which can be used to fit the available daily counts of infection cases.

The data used for the predictions pertain to the entire country, and for the eight most affected  states in the country including Maharashtra, Tamil Nadu, Delhi, Gujarat, Uttar Pradesh, Rajasthan, West Bengal, and Madhya Pradesh.

Talking about the model development Prof. Das said, “The model uses only the daily infection counts available in the public domain without accessing sensitive information pertaining to medical records or contact-tracing data for a large fraction of the population. Despite that, the prediction curves show remarkably good fitting with the past data, and can be used for future planning.”

Prof. Virendra Kumar Tewari, Director IIT Kharagpur said, “People have been living in an uncertain black box without the knowledge about which way life is going to turn and how to plan their activities. A study like this based on a clear statistical model will enable them to understand and plan their wayforward. The model though experimental could prove to be helpful in planning our academic semester and policy matters related to the Institute and the campus under the current circumstances.” 

However, the predictions for the future change quite rapidly with time. There are several potential factors for this such as different mobility patterns of Indian people in different phases of lockdown, large-scale migration of laborers, change in diagnostic facilities, evolution of the coronavirus, and so on. These are well beyond the control of the logistic model or any other currently known prediction model for that matter. 

Although the implementation fails to generate stable and reliable predictions at the moment, the trend clearly reveals that the disease is going to stay in the country for many more months. 

“Our study indicates that India is yet to achieve a steady pattern in the spread of the disease. It is unlikely to get rid of COVID-19 before the end of September 2020,” remarked Prof. Das.

He further added, “This does not leave us in a region of comfort, but the reality has to be accepted, and appropriate plans must be chalked out to address all the issues associated with the outbreak of the pandemic.”

Download Report


Contacts: 

Project Information: abhij@cse.iitkgp.ac.inInstitute Related: Prof. B N Singh, registrar@hijli.iitkgp.ac.inMedia Outreach: Shreyoshi Ghosh, shreyoshi@adm.iitkgp.ac.inMore News:: https://kgpchronicle.iitkgp.ac.in/Follow IIT Kharagpur on Social Media: Facebook:@IIT.Kgp; Twitter: @IITKgp; Instagram: @iit.kgp

By Shreyoshi Ghosh

12 Comments

  • Asik Hossain -

    How you expect that cases will reach about 7.5 lakh around. I thik it will reach minimum of 13 lakh in November.
    Your RSME is very high that’s why I think RSME is not suitable for error calculation in logistic regression model.

  • Kaushik Shekhar -

    Do the data used by this prediction model is reliable??
    Gov just adjust data of previous month to next month.

  • This is really a nice article to get useful information about the COVID-19 outbreak. However, I would really appreciate the author if he could suggest some articles/posts about the wayout in dealing with such a unique outbreak. Thanks in advance! Please keep writing more stuff like this.

  • Ramakrushna Mishra -

    You have estimated the peak will be on 19/06/2020. But it is still increasing. May be you have not taken community spread into account.

    • Shreyoshi Ghosh -

      Contact researcher for further discussion and share your ideas.

  • Rajiv Sharma -

    In between 5th September to 23september I hope it will come in control after that it will continue but with very low potential..and then ends in india. I think…

Leave a Reply

Your email address will not be published.

Related Posts