Scientists see flaws in SUTRA's approach to modelling pandemic

Excellent study material for all civil services aspirants - begin learning - Kar ke dikhayenge!

Scientists see flaws in SUTRA's approach to modelling pandemic

---

Read more on - Polity | Economy | Schemes | S&T | Environment

---

• SUTRA model fails: As India's daily infection rate touched 4 lakh, scientists began questioning the govt-backed SUTRA model, built to study the rise and ebb of the pandemic. In fact, it is being held responsible for misleading everyone into believing no catastrophic second wave is likely at all!
• Insider view: Officials are saying now that SUTRA - Covid supermodel - was important, but neither unique nor determining. 

1. SUTRA group gave inputs to NITI Aayog's Dr VK Paul, who heads the committee taking various inputs from modellers
2. Practical measures were taken once these inputs were taken, and worst case scenarios studied (by National Empowered Group on Vaccines, and Dr Paul's team)
3. Everyone failed as the surge was many times more than expected!
4. On May 2, SUTRA's PIB release said that "a second wave would peak in third week of April and not exceed 1 lakh cases"  (a total failure)

• SUTRA: The "Susceptible, Undetected, Tested (positive) and Removed Approach" model announced in October 2020 that India had crossed the peak (of first wave). 

1. It predicted that maximum Covid burden would be 1.06 crore, in Jan 2021, and just 50000 active cases
2. Other scientists have analysed SUTRA predictions and found many to be totally out of bounds of actual numbers
3. Experts found IIT Kanpur Prof. Manindra Agrawal's (mathematician) tweets proof of the extremely wrong predictions by this "Covid supermodel", and claimed that it was surely a disaster waiting to happen if the govt. was primarily relying on this one "supermodel", which was totally faulty

• Maths not a saviour: The government chose to rely more on mathematical modelling experts, and not actual epidemiologists and public health experts. 

1. Prof Agrawal's defence is that SUTRA model chose a "data centric approach" (and not an epidemiological approach)
2. It used certain "constants" to predict the future case load and possible peaks
3. The equations proved wrong, ultimately, and gave poor results. Basically, very poor use of real world data was done.

• Excessive force-fitting: Other experts say that no model, without ample amount of external inputs from real world data, can reliably predict such waves. The SUTRA model was too complex, with too many parameters, and the professors simply recalibrated those parameters everytime predictions went wrong. 

1. More parameters directly mean overfitting. An expert can fit anything by changing parameters regularly, no big deal!
2. Prof M Agrawal said that the SUTRA model did not anticipate the virus mutants at all and hence 'constants' went wrong
3. Agrawal said actually the model also relied on an ICMR May survey that said 0.73% of India may have been infected at that time (so the actual infected population must have been much lower, results were wrong). So the model assumed that by Jan 2021 more than 50% population was immune (wrong). And many immune people lost immunity with time.

• Positive contribution: Agrawal claims that as per SUTRA, the factor by which reported cases differ from actual ones is a parameter in the model, that can be estimated from just reported govt. data. And that's a big contribution.
• Summary: All said and done, the technical guidance that government of India should have got from experts, was way off the mark. The result is widespread death and debilitation now, across India.

Read more on - Polity | Economy | Schemes | S&T | Environment

Join UPSC Self Prep course | Online Course (PT Gurukul)  | Covid Special Course | Join PT courses

* Content sourced from free internet sources (publications, PIB site, international sites, etc.). Take your own subscriptions. Copyrights acknowledged.