BENGALURU: Researchers at the Indian Institute of Science (IISc) and the Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR) have developed a predictive model for spread of SARS-COV-2 — which causes Covid-19 — that gives state-wise projections for the next four weeks, starting Thursday.
They argue that at a time of a pandemic, it is imperative that predictive models specific to India are designed and developed. The model, developed by professor Santosh Ansumali from JNCASR and Aloke Kumar from IISc, offers a state-wise analysis, throwing up several features of the pandemic wave in India.
A major take away from the projections, Aloke says is the fact that the pandemic wave is likely to hit different states at different time-frames. “Our estimations show that different states will see different parts of the Covid-19 wave at different time-frames in the next four weeks. This means that if our states work well together, they can deal with the pandemic wave in an efficient manner by sharing resources and manpower if need be,” Aloke said.
So, state-wise projection will allow for the most optimal resource planning, the researchers argue. “There have been reports that the infections are under reported. There have been all kinds of projections with some claiming that 10 crore could get affected, while some others claim that 100 crore could get affected. So we created this model that attempts to give a more accurate projection. We wanted an upper bound for which the system must be prepared for. That different states will be hit at different times, resources can be planned state-wise,” Ansumali said.
He further said that such projections are important to plan utilisation of resources and procurement of the same. “Every critically ill person means one ventilator. From what we understand this is an area the authorities need to be prepared for, while oxygen too will become critical,” Ansumali said.
The researchers have used a heuristic algorithm to develop the model and Ansumali said that the model uses the number of deaths reported and works backwards to understand the number of infected and then uses this data to make the projections.
“To address the concerns that there could be under reporting of cases in India, we would like to point out that yes, some cases could have fallen through the cracks. According to us, the number of infected cases now could be around 1,500 or at best 1,800 as opposed to the numbers given out. But it certainly cannot be 10,000 or so as some are claiming,” Ansumali pointed out.
Stating that the team would next try and build a district-wise model, the researchers argued that a state-level granularity of predictions is now a necessity.
“The lockdown has pushed India into clusters of sub-populations, which are expected to follow trend-lines already witnessed in other nations,” they said, adding that their model captures the essential trend-lines from open-source data from other nations.