It might have had an outsized function in creating the notion {that a} catastrophic second wave is unlikely
With shut to 4,00,000 instances being added on daily basis, questions are being raised by a variety of scientists on whether or not a government-backed model, referred to as SUTRA, to forecast the rise and ebb of the coronavirus epidemic, might have had an outsized function in creating the notion {that a} catastrophic second wave of the pandemic was unlikely in India.
An official related with the COVID-19 administration train mentioned, on situation of anonymity, the SUTRA model enter was “an important one, but not unique or determining”.
The SUTRA group offered their views to Dr. VK Paul, who chaired a committee that acquired inputs from a number of modellers and sources. “The worst case predictions from this ensemble were used by the National Empowered Group on Vaccines and the groups headed by Dr. Paul to take measures. However, the surge was several times what any of the modellers had predicted.”
On May 2, the SUTRA group put out a press release, carried by the Press Information Bureau, that the federal government had solicited their inputs the place they mentioned a “second wave” would peak by the third week of April and keep round 1 lakh instances. “Clearly the model predictions in this instance were incorrect,” they famous.
Past its peak
SUTRA (Susceptible, Undetected, Tested (optimistic), and Removed Approach) first got here into public consideration when considered one of its professional members introduced in October that India was “past its peak”. After new instances reached 97,000 a day in September, there was a gentle decline and one of many scientists related to the model growth, M Vidyasagar, mentioned at a press convention then that the model confirmed the COVID burden was anticipated to be capped at 10.6 million symptomatic infections by early 2021, with lower than 50,000 energetic instances from December. In October, at the moment, there have been 7.4 million confirmed instances of which about 7,80,000 have been energetic infections.
Computational biologist, Mukund Thattai, of the National Centre for Biological Sciences, Bengaluru, in a Twitter thread summarised cases of the SUTRA forecasts being far out of bounds of the particular case load. “The so-called Covid ‘supermodel’ commissioned by the Govt of India is fundamentally flawed,” he tweeted. “Based on Prof. Agrawal’s own posts, it was quite clear that the predictions of the SUTRA model were too variable to guide government policy. Many models got things wrong but the question is why the government continued to rely on this model, than consult epidemiologists and public health experts,” he advised The Hindu.
In an e mail to The Hindu, Mr. Agrawal admitted that the model, which had a number of functions, didn’t work effectively on a metric of “predicting the future under different scenarios”.
He mentioned in contrast to many epidemiological fashions that extrapolated instances based mostly on the prevailing variety of instances, the behaviour of the virus and method of unfold, the SUTRA model selected a “data centric approach”. The equation that gave out estimates of what the variety of future infections could be and the chance of when a peak may happen, wanted sure ‘constants’. These numbers saved altering and their values relied on the variety of infections being reported at numerous intervals. However, the equation couldn’t inform when a continuing modified. A speedy acceleration of instances couldn’t be predicted prematurely.
Too many parameters
Rahul Siddharthan, a computational biologist on the Institute of Mathematical Sciences, in an e mail mentioned no model, with out exterior enter from real-world information, may have predicted the second wave. However, the SUTRA model was problematic because it relied on too many parameters, and recalibrated these parameters each time its predictions “broke down”. “The more parameters you have, the more you are in danger of ‘overfitting’. You can fit any curve over a short time window with 3 or 4 parameters. If you keep resetting those parameters, you can literally fit anything.”
Explaining the model’s working, that appears to work higher when the variety of instances reported on a regular basis don’t fluctuate an excessive amount of, the principle causes for it not gauging an impending, exponential rise have been, in accordance to Mr. Agrawal, {that a} quantity, referred to as beta indicating contact between individuals and populations went improper. “We assumed it can at best go up to pre-lockdown value. However, it went well above that due to new strains of virus.”
Further the model was ‘calibrated’ incorrectly.” The model relied on a serosurvey performed by the ICMR in May that mentioned 0.73% of India’s inhabitants might have been contaminated at the moment. “ I have strong reasons to believe now that the results of first survey were not correct (actual infected population was much lower than reported). This calibration led our model to conclusion that more than 50% population was immune by January. In addition, there is also possibility that a good percentage of immune population lost immunity with time.”
Unlike epidemiologists who wanted “accurate data,” within the SUTRA strategy, the issue by which reported instances differ from precise ones is a parameter within the model that might be estimated from simply reported information, (covid19india.org), in accordance to Mr. Agrawal. “I understand it may appear a bit mysterious, but the math shows how. This, in fact, is one of our central contributions,” he advised The Hindu. This has been described in a preprint analysis paper that has been accessible on-line since January.
The modelling examine referred to as the “COVID-19 India National Supermodel” was the results of evaluation by an professional committee consisting of mathematicians and epidemiologists — although in a analysis paper explaining how the model labored, there are three authors: Manindra Agrawal, a professor of laptop science on the Indian Institute of Technology, Kanpur; M. Vidyasagar, a veteran professor {of electrical} engineering on the Indian Institute of Technology, Hyderabad and Madhuri Kanitkar, paediatric nephrologist and Deputy Chief, Integrated Defence Staff (Medical) within the Army.
While many teams of epidemiologists, illness specialists and teams of mathematicians had developed a number of sorts of fashions to predict the result of the pandemic, this group was facilitated by the Department of Science and Technology and was the one one amongst a number of forecast teams, whose numbers have been relayed utilizing the federal government’s publicity channels.
Until February, the model appeared roughly proper, the curve was declining and as of mid-February whereas 10,000-12,000 new instances have been added each day, the general numbers have been shut to 10 million.
Overall caseload
In an interview with this newspaper revealed on February 27, Mr. Agrawal asserted {that a} “second wave was unlikely” although a slight pickup — to about 15,000 instances a day — had begun. India’s total caseload wouldn’t lengthen past mid-March and solely 300,000-5,00,000 new confirmed infections over the subsequent 10 weeks have been anticipated which might convey the general load to 11.3 or 11.5 million infections by April 2021. This was premised partly on 60% of the inhabitants having been uncovered to the virus.
On April 2 after India was within the throes of the second wave, he advised the PTI that the brand new instances would ‘peak’ by April 15-20 — in keeping with the SUTRA workforce’s public assertion.
On April 23, he once more reported a brand new peak, at May 11-15 with 3.3-3.5 million whole ‘active’ instances and decline steeply by the tip of May. India is presently at about 3.4 million energetic instances.
Gautam Menon, a modeller and Professor, Ashoka University, Sonepat, Haryana, who additionally labored on estimating the unfold of COVID-19 disagreed with Mr. Agrawal’s strategy, on the grounds that it was “somewhat simplistic and insufficiently informed by epidemiological data and expertise”.
At finest, Mr. Agrawal’s model might be used together with an ‘ensemble’ — the place outcomes from numerous situations have been grouped. “The use of machine learning to forecast epidemic spread is a relatively recent advance. Some of those models do quite well. But the problems with those methods is that you can’t really figure out what they are doing and how sensitive they are to simply bad data. I would use those models, if we had them, along with an ensemble of other models, but would not repose utter faith in them.”
The SUTRA mode’s elission of the significance of the particular behaviour of the virus: that some individuals have been larger transmitters of the virus than others (say a barber or a receptionist greater than somebody who labored from dwelling), its lack of accounting for social or geographic heterogeneity and it not stratifying the inhabitants by age because it didn’t account for contacts between completely different age teams additionally undermined its validity.
New variants
Mr. Agrawal — who now recurrently tweets on the evolution of the pandemic in States and districts — responded that new variants confirmed up in SUTRA model as enhance in worth of parameter referred to as ‘beta’ (that estimated contact price). “As far as the model is concerned, it is observing changes in parameter values. It does not care about what is the reason behind the change. And computing new beta value is good enough for the model to predict the new trajectory well.”
He conceded {that a} mixture of excellent epidemiologists, data-centric modelling like SUTRA and time-series fashions labored finest. “Time-series based predictions are good at detecting changes in data patterns. So they can flag, early on, phase changes. SUTRA-type data-centric models can explain the past very well [and in studying what was the effect of policy actions, leading to better knowledge-base for future]. They are also very good at predicting future trajectory assuming phase does not change.”
In 2002, Mr. Agrawal and two of his college students developed a mathematical take a look at referred to as AKS primality that would effectively decide in case you may inform an enormous quantity was prime that gained them international accolades. He used a pc science strategy to resolve an issue of pure math. “This is the second time I am entering a domain as a complete outsider. First was when I proved primality theorem. Mathematicians all over the world welcomed a computer scientist in their fold, and in fact went out of their way to celebrate it. Our paper was not written in standard math style, however, experts quickly shut down anyone who questioned the presentation or minor errors in the paper. In contrast, I am experiencing a hostile reaction from epidemiologists, at least in India.”
