NEW DELHI: Days before Koyambedu market, a wholesale hub in Chennai, emerged as a major hotspot leading to a spike of over 200 new Covid-19 cases, a team of researchers at IIT-Madras had submitted reports to state government officials based on their work on mobility data mapped from Facebook users. This was then integrated with data from the crowdsourced portal of covid19india.org.

These reports, which track lockdown violations and mass movements in the state, shared from March 22 with the state government officials, show how lockdown 3.0 is witnessing increased mobility, particularly in Chennai, especially during weekends.

The IIT-Madras team mapped the mobility data of Facebook users in every district in Tamil Nadu, and came up with an analysis on how lockdown compliance in Chennai was far less than in other districts. They showed pockets in North Chennai, areas in and around Central station, Koyembedu market and some fishing hamlets as areas with large-scale movement in spurts.

According to Balaraman Ravindran, professor at the Department of Computer Science and Engineering, who heads the Robert Bosch Centre for Data Science and AI, at IIT-Madras, the data helped the team draw a co-relation between the areas that complied with the lockdown and the spread of infection there.

The location feature of Facebook lets anonymised data get pulled from smartphones and other devices with GPS to generate a map that provides area-level information about distance and movement of people, Ravindran said. On Wednesday, TN recorded 771 cases, with 324 in Chennai alone.

On Monday, reports showed a 17% increase in mobility in the state, compared to during the April 26-29 lockdown and a 7% increase from the April 14-25 lockdown (2.0). Mobility on Monday showed a 16% increase from April 26-29 lockdown but a 15% decrease from lockdown 2.0. Mobility decreased in Chennai after it became a red zone, but picked up soon after, the reports show.

Ravindran said there are two data points that his team managed to get from Facebook –– which areas saw large-scale movement during a lockdown, and how many travelled from a certain part of the state to another at a particular time.

“The base date was set as March 14. So, there was another level of masking, as every detail about the movement was in relation to the movement recorded on March 14. Areas had to be mapped according to the tile set by Facebook. That is how we recovered the movement in every area corresponding to the tile. To know the infection spread, we mapped it with data we obtained from covid19india.org.”

“In India, around 50 million people have the location feature turned on. We do not get this information directly from Facebook. What we get is how many people were in a certain part of the city at a particular time,” he added.

According to researchers, mobility data reports can “help policymakers better monitor compliance and make informed adjustments as the situation evolves”.

The findings show that while some people continued to flout public health guidelines and lockdown protocol till March 27, there was more adherence after that.

“Erode was interesting because mobility was quite high in the early days of lockdown but violations reduced as the days went by. Cases also went down,” Raveendran said. Compliance to the lockdown nationally was over 60%, which was better than most countries.

The team has mapped mobility and infection spread in Chennai, Cuddalore, Ariyalur, Sivagangai, Perambalur, Villupuram and Kanchipuram, and others.

Increase in mobility was observed during weekends, particularly in Chennai, Kancheepuram and Thiruvallur. Overall, mobility has increased in lockdown 3.0, the reports showed, but inter-district movement fell after April 30 for a few days.

“We do not claim to capture the movement of migrant workers as most of them may not have smartphones. Surprisingly, we got a lot of data mapped from rural Tamil Nadu because people use Facebook even in villages,” Ravindran said.

Tamil Nadu health secretary Beela Rajesh confirmed that the government control room was getting frequent reports from IIT-Madras and Facebook. “Data is an important part of our Covid-19 surveillance. We have data committees and control rooms monitoring reports we get from various sources.”

ET has learnt that Facebook is working with researchers in Odisha and Andhra, with IIT-Tirupati on similar projects and helping state governments know how much of the lockdown protocol is being adhered to.

“Disease prevention maps have helped organisations respond to health emergencies for over a year. In the coronavirus context, researchers and non-profits can use the maps, built with aggregated data that people opt in to share, to understand and help combat the virus spread,” a Facebook spokesperson told ET.

Source Article