This is a thread about projections in India and my recent conversation with @karanthapar_in published today at the @thewire_in.

Please recognize that the reality on the ground is changing everyday. Models are wrinkled with assumptions and we have been wrong in the past.
1/n

Deaths are the best predictor of the course of the pandemic and the IHME model is predicting daily deaths to be peaking in mid-May at 5500 deaths/day at the peak. The estimated # infections (reported +unreported) peaks in early May at about 4.5M. 2/n https://covid19.healthdata.org/india?view=daily-deaths&tab=trend

Our models with references available at http://covind19.org  predicting a peak in number of REPORTED cases at 8-10 lakhs (if we could test that many). That is nearly a million cases a day. The # daily deaths at the peak in the second half of May is predicted around 4500. 3/n

There is underreporting of cases in India and everywhere in the world. We estimate this factor between 10-20 in all our work. IHME current estimate for the daily underreporting factor is around 15. That tells us only 3 lakh cases are reported but the situation is much worse. 4/n

With strict regional lockdowns, mask mandates, banning large gatherings, restricting inter-state mobility, increasing vaccination, we can try to bring these projections down. A monumental task lies ahead of us. Colossal mistakes have been made by ignoring signs of this surge. 5/n

Even if the peak is reached in May, it will take a while for the cases and deaths to come down to a level where we can have confidence in resuming normal life. We can only do that if there is a high alert public health system in action which involves genomic sequencing. 6/n

Right now regardless of the models, we know a tidal wave is hitting us and we have picket fences to beat it. We need international coalition, every possible help to increase oxygen supply and produce vaccines. Vaccines work and can save lives. 7/n

As a humanitarian data scientist, I feel there is no social safety net for Indian citizens now, everyone is fighting for their own survival and trying to protect their loved ones. This is hard to watch. Discussing models and projections simulated in the computer seems futile. 8/n

Still as. a data scientist, I felt the moral obligation to state the projections I have at hand when asked. I want to err here and do not want to cause any panic. We all have to do our part as this crisis has the hallmark of becoming a catastrophe if we do not do our best. 9/n

This will not be the last wave, this will not be the last variant, we need cogent and clear public health action plans. This may cost now but will save us from astronomic economic and social costs in the future. We are losing countless human lives that could have been saved. 10/n

Finally, data suppression harms any planning in terms of resource allocation. Knowing the actual reality is always better for the public, for policymakers and for scientists. I am praying for a miraculous recovery for India as we all prepare for a covid-adaptive future. n/n

And here is my conversation with Karan Thapar which this thread is based on.