Rethinking Coronavirus: The Artificial Intelligence Solution

Credit: Centers for Disease Control and Prevention

How AI Can Preemptively Prevent Coronavirus Outbreaks

What was once a meme in February has become a global pandemic just a month later. SARS-CoV-2 or Coronavirus certainly needs no introduction as its exponential spread has united the world in isolation. To arrest the spread of Coronavirus, which has now infiltrated over 150 countries, humans have turned to technology to track and limit its spread. Several prominent approaches have stood out including tracking cell phone data and charting the virus’s spread on real-time interactive maps, and while these methods can adequately highlight areas of high risk, they fail to give us insight into future movements of the virus. As we scramble to understand all the intricacies of the Coronavirus, it is important to realize that we cannot rely on most technology to fill in the gaps for us. However, there is one specific technology capable of learning independently and drawing conclusions from seemingly random data. Artificial Intelligence is the key to finally getting in front of Coronavirus and preventing damage rather than assessing the damage.

Cell Phone GPS Tracking Data Solution

Currently, there are two methods to use cell phone GPS data to alert people who could have come into contact with a Coronavirus disease carrier. The first is an app that tracks whoever has been in close proximity with someone flagged for having Coronavirus and automatically texts a warning to those at risk. FluPhone, an app developed by scientists in Cambridge, assesses who is at risk using Bluetooth and other wireless signals as a proxy for interactions between people, but the problem with it and other apps of this nature is that they require users to self-report flu-like symptoms or other factors that may categorize them as a risk. Since this process relies on user input, sick people who do not make a note of their condition will not be categorized as risky interactions and will, therefore, compromise the functionality of the app. Additionally, using apps to track disease is only effective if the majority of the population uses it, and this was not the case with FluPhone as fewer than 1% of people in Cambridge downloaded the app.¹

The second way cell-phone tracking can be utilized is by allowing governments to access secret phone data in order to retrace the movements and interactions of people who have contracted the Coronavirus. While this method solves the problem of variability, it brings up issues about privacy, since government cell-phone tracking is typically only used as a counterterrorism measure. As a result, the only country that tried implementing this system -Israel- was unsuccessful in spite of officials claiming that the data would be solely used for tracking Coronavirus.²

Even if an ethical and automated phone tracking system was set in place, cell phone tracking at its best can only paint a crude picture of the Coronavirus distribution since most phones can only determine their user’s position with an accuracy between 7 and 13 meters in urban areas, according to a study conducted by Krista Merry and her team.³Additionally, assessing risk by tracking intersections of people can result in false alarms if two people are in proximity for a split second or missed detection if one person occupies the same space as an infected person after the infected person leaves. For this reason, we have turned to real-time interactive models in order to follow the disease as accurately as possible.

Real-Time Modeling Solution

Navigating through a real-time model detailing the spread of Coronavirus is one of the best ways to put into perspective the magnitude of this current pandemic. Not only do the most comprehensive models such as the one developed by Johns Hopkins illustrate the location and frequency of the disease on a world map, but they also track the rate that the virus is being spread daily.

Picture of Johns Hopkins Coronavirus Model
Credit: Johns Hopkins

This technology is incredibly user-friendly and a reservoir of knowledge for Coronavirus researchers, but the information that these models generate still does not tell us where exactly Coronavirus will flare up next. Using past and present data from these models to predict future outbreaks will not be overly insightful because the model will only be able to predict disease spread in places where the disease has already wreaked havoc. Additionally, we do not need a model to tell us that places with higher population and population densities are at higher risk. Finally, relying on the real-time spread of the disease to predict future outbreaks will lead to oversimplified predictions as the nature of the disease spread will constantly be changing with new governmental actions and scientific breakthroughs like vaccines. Because updating the spread of Coronavirus in real-time is not fast enough to save everyone from the disease, we must utilize predictive models that do not only rely on disease statistics in order to gain the edge over Coronavirus. The task of combining fluctuating and vast data inputs with the disease statistics as well as filling the gaps in our understanding of the virus will require artificial intelligence.

Data
Photo by Markus Spiske on Unsplash

Artificial Intelligence Solution

Artificial Intelligence allows predictive models to take in a multitude of variables such as fatalities, confirmed cases, maps of population densities and demographics, test results, tracing contacts of infected people, traveler flows and migration, availability of health-care services, drug stockpiles and other factors.⁴ In other words, AI builds upon real-time model predictions with patterns that it learns from other inputted data. This is possible because of AI’s ability to make sense of unstructured data, tying together tweets, news headlines, and disease statistics as well as other variables in order to generate new insightful patterns. This is pivotal because one major problem with the Coronavirus is the disparity between the confirmed cases and the number of people actually infected (The United States’ inefficient Coronavirus screening process is a topic for another day). A combination of AI and machine learning can overcome this because Artificial Intelligence does not make predictions solely based on recorded data or people who have shown symptoms. One example of AI that is infused with outside the box thinking is a Canadian AI company known as BlueDot. Access to global airline ticketing data allowed BlueDot to predict the outbreak in Wuhan, China three days before government officials finally started warning people, and BlueDot was also able to predict the spread of Coronavirus from Wuhan to Bangkok, Seoul, Taipei, and Tokyo.⁵

Lotus Labs Solution

With expertise in emerging technology and wanting to step up to the challenge, Lotus Labs is in the process of developing two solutions. The first solution is the creation of a chatbot that can readily answer questions related to COVID-19. This chatbot will be able to inform users about their Coronavirus risk levels as well as keeping up with ongoing developments such as mandated quarantines and travel bans. The second is the development of our own AI model that will be able to predict new epicenters of Coronavirus outbreaks in the United States. Our model’s ability to predict where future virus flare-ups will arise will help preemptively warn people living in areas of future risk to be prepared for isolation and take extra precautions before the virus even reaches them. Up until now, Coronavirus has put us on the defense, but now with the help of AI, it is time to attack Coronavirus before it can spread further. We look forward to sharing our research and insights shortly.


[1]: Knight, Will. “Phones Could Track the Spread of Covid-19. Is It a Good Idea?” Wired, Conde Nast, 29 Mar. 2020, www.wired.com/story/phones-track-spread-covid19-good-idea/.

[2]: Halbfinger, David M. “To Track Coronavirus, Israel Moves to Tap Secret Trove of Cellphone Data.” The New York Times, The New York Times, 16 Mar. 2020, www.nytimes.com/2020/03/16/world/middleeast/israel-coronavirus-cellphone-tracking.html.

[3]: Merry, Krista, and Pete Bettinger. “Smartphone GPS accuracy study in an urban environment.” PloS one 14, no. 7 (2019).

[4]: Loten, Angus. “Scientists Crunch Data to Predict How Many People Will Get Coronavirus.” The Wall Street Journal, Dow Jones & Company, 17 Mar. 2020, www.wsj.com/articles/scientists-crunch-data-to-predict-how-many-people-will-get-coronavirus-11584479851.

[5]: Halbfinger, David M. “To Track Coronavirus, Israel Moves to Tap Secret Trove of Cellphone Data.” The New York Times, The New York Times, 16 Mar. 2020, www.nytimes.com/2020/03/16/world/middleeast/israel-coronavirus-cellphone-tracking.html.