Start-up predicts next virus hotspot with social media and mobility data

"This is the brother of the butterfly effect theory, which was about prediction for prevention," said Cobwebs president Omri Timianker. "Our methodology is prediction for decision."

By EYTAN HALON
(photo credit: CHRISTINA HORBAT)
(photo credit: CHRISTINA HORBAT)
In the months-long battle against the coronavirus outbreak, it is often a matter of all hands on deck. For many states, including Israel, that means intelligence agency technologies have played a vital role in monitoring the spread of the virus.
At Cobwebs Technologies, founded in 2015 by three alumni of elite IDF units, cyber experts believe that they can accurately predict the future spread of the virus through publicly accessible information alone.
The Herzliya-based company's proprietary web intelligence system autonomously scours the internet for masses of relevant data, signals and patterns, usually assisting customers in the national security, law enforcement and private sectors in some 35 countries worldwide.
Now, Cobwebs has adapted its solution to predict mobility trends from coronavirus hotspots, promising vital insights to policymakers facing critical coronavirus containment decisions. As countries brace themselves for a second wave of the coronavirus, Cobwebs is currently implementing or testing its solution in numerous Asian Pacific and European countries, and several US states.
"This is the brother of the butterfly effect theory, which was about prediction for prevention," said Cobwebs president Omri Timianker. "Our methodology is prediction for decision."
Cobwebs' solution combines three enormous layers of publicly available data to predict trends and the next possible coronavirus hotspot: web intelligence (WEBINT) based on social media posts discussing testing, symptoms, and sick relatives; open-source location data to identify mobility patterns; and local proprietary data from municipal authorities.
The company has also announced the appointment of Maj.-Gen. (res.) Roni Numa, who was tasked last month by the government with preventing the continued spread of the coronavirus in Bnei Brak, to manage their prediction platform.
"The combination of mobility patterns, web intelligence and methodology can improve a decision-maker's ability to predict and manage their exit strategy," said Timianker.
"What we are focusing on is trends. I don't care who is giving the signals, we are just trying to catch the signals and see how it spreads - it is non-intrusive. The day after the coronavirus, I still do not know who the person is, but I have statistics of potential carriers from high-risk areas to another area."
If there is a high volume of potential coronavirus carriers in Bnei Brak and a trend of Bnei Brak residents traveling to Beersheba is identified, he said, it is then possible to manage and mitigate the likely outbreak in Beersheba ahead of the expected outbreak in the city one week later.
On a local level, if many residents of the same neighborhood with a high level of infection shop at the same mall, the mall can be designated as a potential infection area. Authorities can then take action, either by ensuring greater social distancing in the mall or setting different hours of shopping for different neighborhoods.
In addition to its Herzliya office, Cobwebs has offices in Singapore and New York. The office in Singapore, Timianker said, enabled the company to gain an early insight into both international lockdown measures and domestic exit strategies.
"We wanted to examine different countries in different stages of the outbreak. The most important part of developing a solution is seeing different angles. Our ability to see Europe, Israel, Asia-Pacific and the United States, and see what's working or not, is a big advantage," said Timianker.
"By accurately predicting future trends, decision-makers can use this tool to increase transparency and reduce antagonism caused by lack of trust for everyone. I believe that is important when managing a big crisis."


Related Tags
bnei brak