“I have now computed a range of values for peak value and timing and the final numbers should be within this range. Reason for this uncertainly is that the parameter values for the last phase continue to drift,” tweeted Manindra Agrawal of IIT-Kanpur, involved with the national ‘super model’ initiative, while referring to the updates on peak timing and peak value.
“Peak timing: May 14-18 for active infections and May 4-8 for new infections. Peak value: 38-48 lakh for active infections and 3.4-4.4 lakh for new infections,” Agrawal tweeted on Sunday. He had told TOI last week about the importance of such prediction despite the risk of going off the mark, noting that such an exercise through mathematical model was important for preparing policymakers for a proper response mechanism in terms of medical preparedness, supplies and facilities.
Referring to the model, called Sutra, Agrawal had explained that one should not confuse the two different peaks — one of daily ‘new’ cases which are more commonly observed and another of total number of ‘active’ infections which will come roughly 10 days after the peak for ‘new’ cases.
Update on prediction of peak timing and value means the number of ‘active’ cases in India will keep increasing roughly till mid-May before showing a decline. If the current model shows the trend correctly, the mid-May peak will now be around four times higher than the first peak of over 10 lakh ‘active’ cases witnessed on September 17 last year. India’s total ‘active’ caseload on Sunday reached 26,82,751.
On April 1, the model had predicted the peak of ‘active’ cases somewhere between April 15-20 at around 10 lakh — the same level as what the country saw in September last year. These figures were, however, later revised with the model last week predicting the possibility of a peak between May 11-15 with 33-35 lakh ‘active’ infections.
Asked about the reasons for such large variation in prediction which keeps on changing, Agrawal had then told TOI, “The severity (of the Covid-19 spread) has made computations go haywire. We were seeing significant drift in parameter values for India in our model and so the (previous) modelling was not accurate.”
He said the parameter values kept changing due to new data from states and that’s why the peak value kept shifting. A scientific paper on Sutra by three scientists (Manindra Agrawal, Madhuri Kanitkar and Mathukumalli Vidyasagar) claimed to have applied the model to predict progression of the Covid-19 pandemic in several countries.